I am simulating AODV and AOMDV in NS 2 to compare their performance in terms of Packet Delivery Ratio (PDR), Throughput, Delay and energy consumption, in order to propose a better algorithm. The outputs I am getting after running in NS2 through TCL files are totally different from what I see in other researches especially for residual energy. What am I doing wrong?
It is not possible to get the exact possible data or the the exact model as your base. So, there could be some difference, but clarity of method should be made.
Yes, it is normal...
Unless you have the exact model (and all its peripheral data) and the exact version/release of all programs.
Based on previous experience trying to re-generate simulations, sometimes the results are not very accurate and include "human error". Also, while papers do a good job reporting results, the methods are usually not very detailed... (i..e the small change in settings or change in 1 variable is sometimes not reported)...
Yes it is normal the reasons may be
1)you may have accidentally change some parameters
2}mode of powergui block weather continious or discrete
3}if discrete mode step size also mater
Yes you get different result. Because it is not possible to get the exact possible data or the the exact model as your base. So if in your model it gives better result then u can compare. there is no issue with that.
A typical example of differences: one assumes molar mass of water as 18 g/mol, other 18,0152 g/mol and mass balance of the process will differ by approx. 0,1%
It is very difficult to discuss discrepancies when using some standard models and software packages. They may contain hidden assumptions and limitations.
It is not possible to get the exact possible data or the the exact model as your base. So, there could be some difference, but clarity of method should be made.
This is normal. It means that either noise data has been introduced into your data, or the model you are using is not suitable, may be according to the distribution of the data.
Then you have to try other models or modify them. There must be one option that will provide results with an optimized error and a strong correlation coefficient with your experimental data.
- Badges
- Science topic
- Similar topics
- Biological Science
- Kinesiology
- Running