Check out the new USENIX Web site. next up previous
: Downstream Bandwidth estimation in : Potential Bandwidth Estimation : Background


Beacon Delays

In order to estimate the downstream bandwidth from the access point to an end-host, we propose a methodology that allows the end-host to estimate the delays of the periodic Beacon frames sent from an access point. Figure 1 illustrates how beacon frame transmissions are handled at an access point.
Figure 1: Beacon Transmissions at an Access Point
\begin{figure}\begin{center}
\input{epsf}\setlength{\epsfxsize}{2.4in}
\epsfbox{beacon_delay.eps}
\vspace*{-2mm}
\vspace*{-6mm}
\end{center}\end{figure}
As seen from the figure, an access point schedules a Beacon frame every beacon interval (typically, 102.4 ms). The time instant at which the access point schedules the next beacon message is referred to as the Target Beacon Transmission Time ($TBTT$). As per the 802.11 standard, time zero is defined to be a $TBTT$. Given the value of the beacon interval, the end-host knows the exact time instants when beacon messages are scheduled for transmission. Once a beacon message is scheduled, it is transmitted according to the normal frame transmission rules. In this paper, we assume that beacon frames are not prioritized over other frames, as implemented in the APs used in our experiments. Handling beacon frame prioritization is an interesting extension and will be considered in our future work. The time difference between the instant when a beacon message transmission begins (as obtained from the timestamp field of the Beacon frame) and the $TBTT$ yields an estimate of the beacon delay, $T_{B}$, which is the total time spent by a beacon frame at the access point waiting for transmission. Since we assume that beacon frames are not prioritized over other frames, $T_{B}$ provides an estimate of the total queuing delay plus the contention delay that will be experienced by a data frame transmitted by the AP. Note that beacon delays are computed solely based on timestamps provided by the access point and thus, synchronization issues do not arise.

We now proceed to describe how we can use observed beacon delays to estimate the downstream bandwidth from an AP to a mobile host.


Table 1: Notations for the computation of downstream bandwidth
$T$ total delay incurred by a data frame from
an AP
$T_{D}$ delay incurred between the instant
when a data frame is scheduled for
transmission to the instant that the frame
is received at the receiver
$T_{A}$ delay of the $ACK$ frame from the
receiver to the sender
$T_{B}$ total contention delay experienced by a
data frame from the AP
$DATA$, $RTS$, $CTS$ size of the data, RTS, CTS frame
respectively
$R$ data rate at which the sender transmits the
data frame
$Rb$ basic rate at which control frames are
transmitted
$B$ potential bandwidth from the $AP$ to the
end-host




next up previous
: Downstream Bandwidth estimation in : Potential Bandwidth Estimation : Background