Hi all, Some great feedback here. This has been a traditional method of doing site surveys but for some reason it’s only becoming a big issue for us for this year. Naturally, there’s been a historical view to reduce the amount of APs we deploy in these areas for financial reasons and to avoid more than three APs in a single corridor to reduce 2.4GHz co-channel interference.
5GHz runs at full power in these areas; traditionally 5GHz runs a lot hotter than 2.4GHz even with the same RRM settings. Now here’s something super interesting we discovered. Ever since moving to 7.6.130.21 WLC code, we’ve noticed more of these issues come in. We’ve moved a few APs back to 7.6.100.16 and found the problems went away. When looking at the client connection history graphs, we find more 5GHz connections on the older code. I’m tipping there’s something weird with 802.11h, introduced with this code release, that’s causing the clients to drop their power level. Same channels in either case, so we’re not hitting a transmit power restriction. We have two cases we are currently investigating that look similar - three APs in corridor (Cisco 3602I), wireless dropouts seen on Apple OSX devices (MacBook Pro Retina), three brick walls between client and AP and in both cases, we’re told that an iPhone 5 was used to test the signal and was working in both cases. Our APs are running at 20dBm transmit power at 2.4GHz; 17dBm at 5GHz when at full power, especially in these cases I have verified. I’m aware that a “power level 1” can mean different things depending on channel (especially in 5GHz) but these coverage issues have come to us from corridor deployments where 2.4GHz is the only reliable way to connect. I’ll consider capping our Tx power to 17dBm for future deployments and surveys. Tristan > On 8 May 2015, at 10:58 pm, Rogers, Michael J. <[email protected]> wrote: > > Out of curiosity what power level do you run your 5ghz band? > > From: The EDUCAUSE Wireless Issues Constituent Group Listserv > [mailto:[email protected] > <mailto:[email protected]>] On Behalf Of Hinson, Matthew P > Sent: Monday, May 4, 2015 8:02 AM > To: [email protected] > <mailto:[email protected]> > Subject: Re: [WIRELESS-LAN] Running APs at full power: client transmit power > levels low? > > Hi Tristan, > > You definitely want to match the Tx power between clients and APs as close as > you can. Obviously, being education, we have little to no control over the > hardware brought into our environment, so always knowing every device’s Tx > power can be hard. > > Wi-Fi is a two way street. If at all possible, a client and an access point’s > power settings should match. Almost every frame sent to a client must be > acknowledged very soon after, and if the client can’t reliably talk back to > the AP, you’re going to have an unstable or unreliable connection. > > We run our APs around 15-17dBm in the 2.4GHz band depending on the area but > never higher. With the proliferation of mobile devices, that’s about all you > can get away with without causing a mismatch. > > Aerohive had a blog post a while back about the iPhone 5 and its 16dBm output > power in the 2.4GHz band. > http://blogs.aerohive.com/blog/the-network-revolution/apple-iphone-5-wi-fi-specs > > <http://blogs.aerohive.com/blog/the-network-revolution/apple-iphone-5-wi-fi-specs> > > > From: The EDUCAUSE Wireless Issues Constituent Group Listserv > [mailto:[email protected] > <mailto:[email protected]>] On Behalf Of Tristan Gulyas > Sent: Monday, May 4, 2015 3:55 AM > To: [email protected] > <mailto:[email protected]> > Subject: [WIRELESS-LAN] Running APs at full power: client transmit power > levels low? > > Hi all, > > We’ve run into an issue in some of our sparsely covered areas (2.4GHz > coverage optimised, not density optimised) where we have APs in a corridor > style deployment. This is typically found in older buildings which means > we’re dealing with solid brick interior walls. > > These APs are typically running at maximum power levels (typically 3600/3700 > series Cisco radios). > > In one case, we measured the client end (MacBook Pro) as -71dBm with an SNR > of 22; the AP end saw the client with an SNR of 14 and a signal of -81dBm and > connectivity was unreliable. I have seen similar results elsewhere with a > similar deployment model. > > Has anyone else experienced similar issues with corridor style deployments at > full power? > > Cheers, > Tristan > > > Tristan Gulyas > Senior Network Engineer > Network Operations > eSolutions | Monash University > 738 Blackburn Road Clayton 3800 > www.monash.edu <http://www.monash.edu/> | [email protected] > <mailto:[email protected]> > > > > > ********** Participation and subscription information for this EDUCAUSE > Constituent Group discussion list can be found at > http://www.educause.edu/groups/ <http://www.educause.edu/groups/>. Tristan Gulyas Senior Network Engineer Network Operations eSolutions | Monash University 738 Blackburn Road Clayton 3800 www.monash.edu <http://www.monash.edu/> | [email protected] <mailto:[email protected]> ********** Participation and subscription information for this EDUCAUSE Constituent Group discussion list can be found at http://www.educause.edu/groups/.
