7.1 Synchronization On Unloaded Network Lab Report
Chapter
7
Discussion of Test Results
7.1 Synchronization on unloaded networks
First considering the result for test case 1.1, a low average value is revealed, if compared to the accuracy stated for NTP in the literature. This low average should be taken with a grain of salt however, as this is not an absolute value as shortly explained in the results of the tests. Therefore, oscillations between -20 ms and 20 ms would for example result in an average difference of 0 ms, why the average difference cannot solely be used to determine whether an acceptable synchronization was achieved. The average differences of all the tests can thus be seen as somewhat optimistic values, and thereby a bit misleading.
As oscillations are possible, this may also …show more content…
Especially a higher maximum difference
61
62
Discussion of Test Results was achieved, which probably was what affected the standard deviation.
7.2 Synchronization on loaded networks
While both NTP and PTP performed quite well on unloaded networks, as dis- cussed in Section 7.1, it was interesting to test the performance in a more hostile environment. WANs would be an example of a such a hostile environment, where the previously mentioned asymmetric network paths may occur. These asymmetric paths are, as explained in Section 2.3.1, not possible to differentiate from the clock offset. The accuracy of NTP can therefore quickly degrade in these environments.
As explained in Section 7.1, the average difference may be misleading, which is also true in these test cases.
In the first test case, 2.1, the average difference of 28 ms does not exceed the upper threshold of 40 ms, as specified in Section 2.2.1, but it is not far from. Fur- thermore, with a maximum difference of about 50 ms and a standard deviation of about eight milliseconds, there will be many cases of the audio not being perceived as synchronized during this test.
Naturally, both the second and third test case, 2.2 and 2.3, also show a
7
Discussion of Test Results
7.1 Synchronization on unloaded networks
First considering the result for test case 1.1, a low average value is revealed, if compared to the accuracy stated for NTP in the literature. This low average should be taken with a grain of salt however, as this is not an absolute value as shortly explained in the results of the tests. Therefore, oscillations between -20 ms and 20 ms would for example result in an average difference of 0 ms, why the average difference cannot solely be used to determine whether an acceptable synchronization was achieved. The average differences of all the tests can thus be seen as somewhat optimistic values, and thereby a bit misleading.
As oscillations are possible, this may also …show more content…
Especially a higher maximum difference
61
62
Discussion of Test Results was achieved, which probably was what affected the standard deviation.
7.2 Synchronization on loaded networks
While both NTP and PTP performed quite well on unloaded networks, as dis- cussed in Section 7.1, it was interesting to test the performance in a more hostile environment. WANs would be an example of a such a hostile environment, where the previously mentioned asymmetric network paths may occur. These asymmetric paths are, as explained in Section 2.3.1, not possible to differentiate from the clock offset. The accuracy of NTP can therefore quickly degrade in these environments.
As explained in Section 7.1, the average difference may be misleading, which is also true in these test cases.
In the first test case, 2.1, the average difference of 28 ms does not exceed the upper threshold of 40 ms, as specified in Section 2.2.1, but it is not far from. Fur- thermore, with a maximum difference of about 50 ms and a standard deviation of about eight milliseconds, there will be many cases of the audio not being perceived as synchronized during this test.
Naturally, both the second and third test case, 2.2 and 2.3, also show a