Timer Synchronization

Like other wireless network technologies, 802.11 depends a great deal on the distribution of timing information to all the nodes. It is especially important in frequency-hopping networks because all stations on the network must change frequency channels in a coordinated pattern. Timing information is also used by the medium reservation mechanisms.

In addition to local station timing, each station in a basic service area maintains a copy of the timing synchronization function (TSF), which is a local timer synchronized with the TSF of every other station in the basic service area. The TSF is based on a 1-MHz clock and "ticks" in microseconds. Beacon frames are used to periodically announce the value of the TSF to other stations in the network. The "now" in a timestamp is when the first bit of the timestamp hits the PHY for transmission.

Infrastructure Timing Synchronization

The ease of power management in an infrastructure network is based on the use of access points as central coordinators for data distribution and power management functions. Timing in infrastructure networks is quite similar. Access points are responsible for maintaining the TSF time, and any stations associated with an access point must simply accept the access point's TSF as valid.

When access points prepare to transmit a Beacon frame, the access point timer is copied into the Beacon's timestamp field. Stations associated with an access point accept the timing value in any received Beacons, but they may add a small offset to the received timing value to account for local processing by the antenna and transceiver. Associated stations maintain local TSF timers so they can miss a Beacon frame and still remain roughly synchronized with the global TSF. The wireless medium is expected to be noisy, and Beacon frames are unacknowledged. Therefore, missing a Beacon here and there is to be expected, and the local TSF timer mitigates against the occasional loss of Beacon frames.

To assist active scanning stations in matching parameters with the BSS, timing values are also distributed in Probe Response frames. When a station finds a network by scanning, it saves the timestamp from the Beacon or Probe Response and the value of the local timer when it was received. To match the local timer to the network timer, a station then takes the timestamp in the received network advertisement and adds the number of microseconds since it was received. Figure 8-19 illustrates this process.

Figure 8-19. Matching the local timer to a network timer

IBSS Timing Synchronization

IBSSs lack a central coordination point, so the Beacon process is distributed. TSF maintenance is a subset of the Beacon generation process. Time is divided into segments equivalent to the interbeacon timing period. Beacon frames are supposed to be transmitted exactly as the beacon interval ends, at the so-called target Beacon transmission time (TBTT). Independent networks take the TBTT as a guideline.

All stations in the IBSS prepare to transmit a Beacon frame at the target time. As it approaches, all other traffic is suspended. Timers for the transmission of frames other than Beacon frames or ATIM frames are stopped and held to clear the medium for the important management traffic. All stations in the IBSS generate a backoff timer for Beacon transmission; the backoff timer is a random delay between zero and twice the minimum contention window for the medium. After the target beacon interval, all stations begin to count the Beacon backoff timer down to zero. If a Beacon is received before the station's transmission time, the pending Beacon transmission is canceled.

In Figure 8-20, each station selects a random delay; station 2 has randomly generated the shortest delay. When station 2's timer expires, it transmits a Beacon, which is received by stations 1 and 3. Both stations 1 and 3 cancel their Beacon transmissions as a result. Because timer synchronization ensures that all stations have synchronized timers, multiple Beacon frames do not pose a problem. Receivers simply process multiple Beacon frames and perform multiple updates to the TSF timer.

Beacon generation interacts closely with power management. Beacon frames must be generated during the active period around each Beacon interval so that all stations are available to process the Beacon. Furthermore, the Beacon sender is not allowed to enter a low-power state until the end of the next active period. The latter rule ensures that at least one station is awake and can respond to probes from new stations scanning to discover networks.

Rules for adopting the received timestamp are more complex in an independent network. No centralized timer exists, so the goal of the standard is to synchronize all timers to the timer of the fastest-running clock in the BSS. When a Beacon is received, the

Figure 8-20. Distributed Beacon generation

timestamp is adjusted for processing delays and compared to the local TSF. The received timestamp updates the local timer only if it is later than the local timer.