Information Technology Reference
In-Depth Information
TaBLe 3.4
Calculated Coefficient by Different Training Set with LSM Algorithm
Training Set Size
10
18
30
β
0
0.1503
0.0528
−1.38407
β
1
−0.0071
0.0778
−0.90331
β
2
0.0016
0.0005
−0.12004
β
3
2.4637
2.2441
−0.43265
β
4
0.4383
0.2818
0.244986
β
5
0.6521
0.8361
0.418736
β
6
−1.4899
−0.9349
−1.22127
β
7
−1.8959
−2.4403
−0.10035
β
8
−0.3229
−0.3236
0.05914
β
9
−0.8456
−0.7946
0.628173
3.5.6.1 Least Squares Method
In order to adopt the LSM, the minimum training set size is 10. The obtained
model was tested with data that is in the training set after the training pro-
cess. The coefficient calculated from the training set is listed in Table 3.4 and
shows the experiment with different results. The experimental result of LSM
is shown in Figures 3.19-3.21 and Table 3.5.
3.5.6.2 Simulated Annealing
A portion of the training data and change in the error for the training process
are show in Table 3.6. The training set size is larger and the mean squared
error root is minimized for these values. This shows that the LS is a good
algorithm to solve the problem.
As seen from the Figures 3.22-3.25, the estimation becomes more accurate
when the training set is larger. There is not much difference with both algo-
rithms when the training set is more than 18, which is the number of the
sample video.
3.5.7 Summary
Based on the aforementioned analysis, we present here how these results can
be further exploited toward developing a video quality prediction method
base on a network status and original frame lost rate. Hence, consider an
unknown video clip in a specific network environment and predict its PEVQ
that better describes its quality after transmission and encode, decode. Thus
the algorithm can successfully predict the video quality. Consequently, the
service provider can predict analytically video service quality that satisfy
specific quality levels at a pre-encoding state.
We will try to make our protocol stack and video communication system
compatible across 3G-324M, Wi-Fi, and H.323. Thus, the video communication
