Wireless Networking → Theory on Optimizing a 802.11 Beacon

If you wish to make your beacon most efficient, choose a prime number from this list:
•1-2 wireless clients: 197 199 211 223 227 229 233 239 241
•3-5 wireless clients: 149 151 157 163 167 173 179 181 191 193
•6+ wireless clients: 97 101 103 107 109 113 127 131 137 139

Reasoning:

The beacon is the rhythm of your AP-controlled network. It performs that function far more frequently than its other famous purpose: to help your network appear in a scan of available wireless networks. The beacon occurs several times a second to perform both of these tasks.

Each channel has a theoretical bandwidth (e.g. 11 Mbps, 54 Mbps).

Each time the beacon transmits, the length of the beacon transmission takes from the available bandwidth of the channel.

If you decrease the time between beacons by too much (reduce the beacon interval), the available bandwidth can become consumed by the increased number of beacons -- which do not contribute to the throughput of your network. And if you only have a couple of wireless clients on your network, most of the beacons will go unanswered because none of your clients have any traffic for the AP.

If you increase the time between beacons by too much (raise the beacon interval), it creates periods of "dead air" as a client holding traffic waits for a beacon. And, if you have multiple stations competing for the attention of the AP, a wide interval between beacons makes it more likely that more than one station will attempt to contact the AP at the same time.

Adjusting the beacon interval to suit your network may increase your bandwidth by balancing the need to open windows for data while avoiding collisions.

Choosing from among a list of prime numbers makes it less likely that your beacons will chirp in sync with the beacons of other networks. Since beacons can collide as well, it makes sense to choose an interval that is unlikely to align with neighboring networks for very long.

If I'm not mistaken, the beacon interval has nothing to do with clients holding traffic until a beacon occurs, with one exception: battery powered clients (PDAs mostly) can be configured to operate in low power consumption mode, meaning they sleep until the next beacon then wake up to see if the access point (or ad-hoc peer) has traffic queued.

Operating without this power saving strategy uses the CSMA/CA principles.

The default beacon interval is 100mSec in most systems. The beacon tranmission time is a tiny fraction of that interval. (Some newer WLAN gear (e.g., Cisco) is placing information in the beacon (a few bytes) to assist in fast handoffs, within the user defined aspects of the 802.11 standard.)

Or I'm all wet.

You're not all wet, but a little damp in a couple of places.

The power saving mode of which you speak uses both the Beacon and the DTIM to tell clients how to figure out when network broadcasts can happen.

So if Beacon Interval=100 and DTIM=10, it can sleep for 100x10 ms (1000 ms) before it has to wake up to hear a beacon.

You're correct about the shortness of a beacon packet, but don't forget that clients have to sync in order to hear data (including the beacon). While the data in a beacon is only a few bytes, there is a period of preamble to help the clients lock on. This extends the impact of the beacon to throughput.

Personally, I'm a little surprised that 100 ms. was picked as a default. I have no science to back up my suspicion that a larger default value might have been more efficient to start with.

From my access point help file
...
Beacon period- The amount of time in Kusecs (one Kusec equals 1,024 microseconds) between radio beacons from the base station to its client stations (laptops, PCs, etc). Beacons are packets sent by an Access Point to synchronize a wireless network.
DTIM interval- A DTIM (Delivery Traffic Indication Message) is a countdown informing clients of the next window for listening to broadcast and multicast messages. This setting, always a multiple of the beacon period, determines how often the beacon contains a traffic indicator map (TIM). The TIM alerts stations in sleep state to stay awake long enough to receive their data frames.
...
If the DTIM is supposed to be a multiple of the beacon period, and you set the beacon period to a prime number, then you can only set the DTIM to 1 or the beacon period?
Both my access point and my network adaptor have settings for the beacon period, would they have to be set to the same value?
Confused

No. DTIM is a multiplier. By calling it a count down, they're saying the same thing differently.

If the beacon is 101 and the DTIM is 1, then the broadcasts can come every 101 ms. (101 x 1). If the beacon is 101 and the DTIM is 2, then the broadcasts can come every 202 ms. (101 x 2).

If the DTIM is 10, it works like this:

• Beacon, Broadcast Eligible, DTIM Counter=10
  
• Beacon, Broadcast Inligible, DTIM Counter=9
  
• Beacon, Broadcast Inligible, DTIM Counter=8
  
• Beacon, Broadcast Inligible, DTIM Counter=7
  
• Beacon, Broadcast Inligible, DTIM Counter=6
  
• Beacon, Broadcast Inligible, DTIM Counter=5
  
• Beacon, Broadcast Inligible, DTIM Counter=4
  
• Beacon, Broadcast Inligible, DTIM Counter=3
  
• Beacon, Broadcast Inligible, DTIM Counter=2
  
• Beacon, Broadcast Inligible, DTIM Counter=1
  
• Beacon, Broadcast Eligible, DTIM Counter=10

Time between broadcast eligibilitiy 101 x 10 = 1010 ms.

DTIM is usually set to 1 since the throughput is bad with longer sleep intervals.