During our
experimental evaluation, we want to control the invoking of the rate switching
algorithm. To emulate a real environment during our controlled experiments,
we used an artificial noise source to alter the condition of the wireless link. In
our experiments we started streaming data for 60 seconds, and then turned on
the noise source causing an instant pulse of noise. After 60 seconds, the noise
source was turned o?® and data continued streaming for another 60 seconds.
We were able to trigger rate switching as seen in Figure 5.
8.2 Results
We analyzed the captures using the encoding process and periodogram estimation
as discussed above. We partitioned the analysis into three parts: the
interval before injecting noise, the interval with noise, and the interval after
injecting noise. During our analysis we used a sampling bin of 2 ms, which
represents a frequency range up to 250 Hz. We also removed the mean from
the sampled signal prior to calculating the discrete-time Fourier transform to
435
Cherita Corbett, John Copeland, and Raheem Beyah
Fig. 5. Linksys card entering and exiting rate switching.
remove the DC bias from the signal. If not removed, the mean gives rise to
significant power at 0Hz.
Our technique produces a spectrum that captures a view of the dynamic
properties of a wireless stream. Because we filtered out all other frame types
during the encoding process to only consider data frames, the spectrum captures
the power or strength of the data transmission rate contained at a particular
frequency.
Pages:
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680