Streaming Variable Bit-Rate Media Streams - Scalable Continuous Media Streaming Systems - page 128

Table 7.1 Comparison of admission complxity between the Optimal Smoothing and the MDR
schedulers

Unsuccessfull Admission # Successfull Admission

Scheduler Comparisons Additions Comparisons Additions

Temporal smoothing
complexity at ρ = 0 . 6 ∗ 0.00 (0.00) 0.00 (0.00) 5782.52 5782.52

complexity at ρ = 0 . 7 ∗ 0.00 (0.00) 0.00 (0.00) 5782.20 5782.20

complexity at ρ = 0 . 8 ∗ 0.19 (0.17) 0.19 (0.17) 5782.04 5782.04

complexity at ρ = 0 . 9 ∗ 5.32 (4.89) 5.32 (4.89) 5782.08 5782.08

MDR scheduler
complexity at ρ = 0 . 6 ∗ 0.00 (0.00) 0.00 (0.00) 1 5782.60

complexity at ρ = 0 . 7 ∗ 0.00 (0.00) 0.00 (0.00) 1 5782.28

complexity at ρ = 0 . 8 ∗ 0.17 (0.17) 0.17 (0.17) 1 5782.11

complexity at ρ = 0 . 9 ∗ 4.83 (4.83) 4.83 (4.83) 1 5782.16

Notes : *Numerical results are measured as the average number of computations required to admit a
client at a given average network utilization ( ρ ), averaged over requests for all 274 videos.
# Numbers in parentheses are the average number of unsuccessful admission tests performed for each
request.

To admit a new stream, network resource reservations are performed according to the gen-
erated transmission schedules on a per-stream basis. Admission test for schedules generated
by optimal smoothing is performed with a round length of one second (cf. Section 7.3.2). Each
simulation run simulates a duration of 3,000 days. We summarize the results in the following
sections.

7.4.1 Admission Complexity
Table 7.1 compares the admission complexity of Optimal Smoothing and the MDR sched-
uler. The simulation results are obtained from counting the average number of computations
required to admit a new client. We separate the computations incurred in unsuccessful and
successful admissions. For unsuccessful admissions, the computation complexity is compa-
rable for the MDR scheduler and the Optimal Smoothing scheduler. By contrast, the MDR
scheduler requires significantly fewer computations than the Optimal Smoothing scheduler for
successful admissions, which dominates the total admission complexity.

7.4.2 Waiting Time versus System Utilization
To evaluate the bandwidth efficiency of the MDR scheduler, we collected the mean and worst-
case client waiting times for both schedulers and plot the results in Figure 7.3 a,b for three
system bandwidth settings. The results show that the MDR scheduler achieves performance
similar to Optimal Smoothing for all three system bandwidth settings and across system uti-
lization from 10% to 90%. This suggests that the MDR scheduler can guarantee VBR video
delivery in mixed-traffic networks with negligible trade-off in latency - the key performance
metric experienced by the end users.