Check out the new USENIX Web site. next up previous
Next: Time-based Regulator Up: Existence of Rate Diversity Previous: Existence of Rate Diversity


Multiple Users during Busy Periods

Figure 5: The fraction of network throughput achieved by the heaviest user at a busy AP during busy 1-second intervals.

We analyzed wireless tcpdump trace of Whittemore, a residential facility in the Dartmouth business school where students were required to own laptops. This data was collected by Kotz et al. over the Spring semester [18] and was made publicly available by Kotz. Unfortunately, the trace data does not contain the data transmission rate used for each frame transmission. Nonetheless, we can identify the busy periods in which an AP is carrying close-to the maximum amount of data, and investigate whether more than one user actively exchange data during congested periods.

Since TCP dominated the traffic, we conservatively define busy or congested intervals as those in which the total data throughput at the AP exceeded 4 Mbps, $80\%$ of the commonly observed TCP saturation throughput when nodes transmit at the maximum data rate and experience a very low loss rate of $1\%$ to $2\%$.

Figure 5 plots the fraction of aggregate throughput achieved during busy 1-second intervals by the heaviest user at an AP at Whittemore on 8 April 2002, a Spring Monday. The heaviest user is one that exchanged the most bytes with the AP. Although the majority of bytes were transferred by one user on average, it is clear that the heaviest user alone rarely saturated the channel. In most 1-second busy intervals, users other than the heaviest user exchanged significant amounts of data.


next up previous
Next: Time-based Regulator Up: Existence of Rate Diversity Previous: Existence of Rate Diversity
Godfrey Tan 2004-05-04