1

Can anyone offer any advice for setting up a wireless network for a small technical conference? (Perhaps ~ 100 attendees, for 1-3 days.)

I've read the resources on this site. For instance, I've read all of the answers to Why is Internet access and Wi-Fi always so terrible at large tech conferences? and the PyCon conference reports. Those are great and had lots of useful information, but they are focused at large conferences, whereas I'm asking about a small conference. Also, many of the answers are from ~ 3 years ago; I don't know if anything has changed since then.

Some specific questions:

  • If I'm bringing my own APs, how should I choose APs? What features should I look for, when I select an AP?

  • How should I configure the APs? Should I give them all the same ESSID (to allow roaming), or assign each a different ESSID based upon its location (to allow users to manually select the AP that is physically closest to them)?

  • How should I assign the channel and power strength of each AP?

  • What's the typical bottleneck that causes poor performance?

I can't afford to spend thousands of dollars or to spend several days before the event. These events are typically hosted at a hotel or conference facility, so I may be forced to work with their existing Internet connection (unless someone has some better ideas for how to get connectivity for a short event).

Here's what I've gathered from the other resources listed above (check whether these lessons sound right to you):

  • Bring your own access points. Bring a bunch, deploy them at physically remote locations, and turn the signal strength way down.

  • The 5 GHz band is much better than the 2.4 Ghz band. It has less interference between channels.

  • 802.11n scales best. Sounds like I can expect up to 30-50 users per AP on 5 GHz 802.11n, vs 10 users per AP on 2.4 GHz (is that right?). One person recommended getting dual-radio APs that can talk 802.11n on both radios (apparently this is tricky?).

  • I'm not too clear on what criteria to use when selecting APs, but it sounds like dual-radio is best, and if possible, it makes deployment easier if they support power-over-Ethernet. I dunno if brand is important (I see Cisco and D-Link mentioned).

D.W.
  • 206
  • 4
  • 13
  • One more thing I forgot to mention: Have someone scan all rooms to locate fake APs. If the conference is large enough someone always seems to want to intercept traffic and snoop passwords. – Hennes Jul 28 '12 at 20:20
  • 100 people is large enough that the existing recommendations for large conferences apply to you, too. – womble Jul 28 '12 at 21:13

2 Answers2

3

If I'm bringing my own APs, how should I choose APs? What features should I look for, when I select an AP?

  • Select APs which support the 5GHz band. (wireless N).
  • For the 2.4GHz band (and I guess you also want to support older devices which only support that): Select APs where you can turn transmission power down (better a lot of APs at low power than a few at high power).
  • Map where you want to place the AP and pre plan which channels, preventing overlap. (Not a problem for N, but you will end up with no more than 3 bandw without overlap for 2.4GHz
  • Try to place the APs somewhere where a human crowd will not block the signal (Humans are mostly water. Water blocks the signal. This is something causes problems during a real event, but not when you test it in an empty room before the even starts)
  • And unless you like a lot of manual work, select an AP which you can manage well from a distance. This often means enterprise APs and a big price tag.


How can I estimate how many APs I will need? How many users can be handled per AP?

Up to 25 user per AP seems to be the fine.


How should I configure the APs? Should I give them all the same ESSID (to allow roaming), or assign each a different ESSID based upon its location (to allow users to manually select the AP that is physically closest to them)?

Same ESSID, unless your users will not walk around and are technical enough to understand how a computer works and why a link gets lost when reconnecting to a new AP. These people seem rare.


How should I assign the channel and power strength of each AP?

That varies per room, layout and neighbouring AP. Everything I read about it seems to indicate that this is non-trivial.


What's the typical bottleneck that causes poor performance?

  • Too many users in a small space (and thus a few barely used AP and a few swamped APs)
  • Too many users on 2.4GHz (that band is crowded. Even when everybody turns other 2.4GHz devices off (e.g. bluetooth)
  • Big downloads. If you have presentations which offer a few iso files or other large files, then also supply some cabled network ports.
  • WPA2 encryption turned on and active for dozens of users at the same time.


I can't afford to spend thousands of dollars or to spend several days before the event. These events are typically hosted at a hotel or conference facility, so I may be forced to work with their existing Internet connection (unless someone has some better ideas for how to get connectivity for a short event).

Now that is a problem. Doing this well is not cheap. Enterprise APs are expensive but there is a reason they are often used.

Falling back to the hotel might be a lot cheaper, but hotels often say "Yes, we have internet. Yes, all guests can use it". Usually that means that the average number of guests can use it for light surfing. A conference might well overload their service. Do not assume that their wireless will work. Do not assume that there is hotel personal with knowledge about the network on the premises.

If you do go for the hotel network: Make sure their technical person is on site that day. Ask keys for the server closet. Etc etc.

Uplink: Make sure you have enough bandwidth. Preferably with multiple separate providers. This is not cheap either.

Here's what I've gathered from the other resources listed above (check whether these lessons sound right to you):

Bring your own access points. Bring a bunch, deploy them at physically remote locations, and turn the signal strength way down.
Agreed

The 5 GHz band is much better than the 2.4 Ghz band. It has less interference between channels.

No. If you select neighbouring channels you also get interference on the 5GHZ bands. But you have way more channels. If you select them wisely you can avoid interference.

One person recommended getting APs that have two 5 GHz radios (not just one).
Agreed

802.11n scales best. Expect up to 30 users per AP on 5 GHz 802.11n, vs 10 users per AP on 2.4 GHz.
That seems sensible. It differs per AP. Some might choke on 15 users, some might work fine with up to 50. But 25 or 30 seems a good average.

Hennes
  • 4,772
  • 1
  • 18
  • 29
1

I don't know if I can answer all your questions, but I'll give it a try:

5 Ghz vs 2.4: If there aren't any thick walls, you are indeed better of with 5 Ghz, it just performs better in an open space.

Another thing you might want to enable is roaming, normally all AP's just need the same SSID and password. You will need at least an overlap of 12.5% in signal for it to work properly, in this way your users will automatically switch AP when walking around without actually noticing.

Also, if possible, use a central server to do DHCP and provide trothling. Use your AP's only as AP's, the last thing you want is your AP's handing out a random a subnet.

Lucas Kauffman
  • 16,818
  • 9
  • 57
  • 92