A similar analysis could be done to isolate the spectral
content of management or control frames within a wireless stream or for a
specific type of frame (i.e., retransmitted frames). Since the rate switching
algorithm is only applied to data frames, we only encoded data frames.
Spectral analysis of the tra?±c trace prior to injecting noise generates a
similar PSD for all three types of RIs as illustrated by Figure 6. The PSD
also reveals that power is concentrated at discrete frequency points. Each RI
has the most prominent peaks at 100Hz and 200Hz, confirming that the RIs
behaved similarly in transmitting data frames when there is no rate switching
(i.e., when the condition of the link is perceived as good).
During the interval in which noise was injected into the wireless environment,
each RI generated a distinct PSD as shown in Figure 7. In contrast
to the noiseless period, we also observed that prominent peaks are scattered
throughout the frequency range, especially at the lower frequencies. For example,
in Figure 7(a) the Lucent RI still has prominent peaks at 100Hz and 200Hz
(that were seen before injecting noise), but new distinctive peaks are found at
the lower frequencies ranges (0-10Hz and 50-60Hz). The spread of prominent
peaks throughout the frequency range indicates that a host is transmitting
data frames at several di?®erent rates. This type of behavior is expected while
the RI is executing its rate switching algorithm.
Pages:
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680