next up previous
Next: Future Work Up: Experimental Results Previous: Scenario 2

Scenario 3

As described in section 3.3, in scenario 3 we endeavored to find the MTBBU vs. the average delay in a manner similar to [1]. The results so far are inconclusive. We believe that the deterministic approximation that we make to find the average delay is either not accurate enough or not computed properly. Also, because we must assume both that tex2html_wrap_inline947 is large compared to tex2html_wrap_inline949 and that tex2html_wrap_inline803 must be relatively large for our approximations to hold, we expect that the MTBBU is extremely high, making it difficult to run simulations long enough to capture good data points. In the plot in figure 8 the disparities are evident. It seems that with a half second shift to the right of the analytical curves there would be agreement for smaller buffer values.

   figure387
Figure 8: We plot MTBBU as a function of buffer size for tex2html_wrap_inline835 , tex2html_wrap_inline1165 sec. tex2html_wrap_inline1167 seconds. This is identical to the case given in [1] We see that the experimental results are similar to those found by Steinbach. The analytical are shifted to the left, however.



Mark Kalman
Tue Mar 13 05:01:37 PST 2001