Sensor Simulation System Article No.4
By
To accomplish this goal using 40 PCs, DS3 has to achieve a data generation and insertion rate of approximately 4:200 aggregated data packages / second on each individual computer.
A series of tests were conducted on a personal computer with a dual core AMD Athlon64-LE-1640 CPU clocked at 2600 Mhz and 2Gb of RAM with a speed of 800 Mhz. We ran tests for five different time intervals: 1 minute, 5 minutes, 10 minutes, 20 minutes and one hour. We allowed for a 5 second setup time for each test and we repeated the test 10 times for each time interval.
The results we have obtained are summarized in table below and show that the current implementation is capable of delivering the amount of data required by our storage engine tests:
|
Time intervals (seconds) |
65 |
305 |
605 |
1205 |
3605 |
|
Average inserts |
332,094.40 |
1,678,241.70 |
3,409,745.90 |
6,751,796.10 |
20,307,480.30 |
|
Average inserts per second (average insertion rate) |
5,109.14 |
5,505.71 |
5,635.94 |
5,603.15 |
5,633.14 |
In order to compare the quality of the simulated data we have charted (the left image) the general power consumption reported by the current Current Cost meter in a 30 minutes test interval in which we precisely monitored household appliance usage. We then used our observations to construct a DS3 simulation scenario that could mimic the real-life experiment. We have also charted our results for the spot household consumption profile.
The results show that, if provided with quality input data, our two stage simulation technique can generate very accurate consumption readings that closely resemble the one produced by real-life hardware monitoring system.
Research Institute e-Austria homepage: www.ieat.ro/
