|
|
In many ways, this has been the busiest week. I feel like the whole area was so new to me that many of the concepts only started making sense to me as late as last week! So I have been working to run some simulations and test my own hypotheses, as well as trying to write them all up in a thorough formal report. As I mentioned last week, it is possible that the number of hops a packet takes should have an affect on the energy consumption. I tried to normalize the data to determine the per packet energy consumption, but this can not be very conclusive data when the number of hops varies. I finally simply plotted the number of packets sent and received against the SPD. In this case, we finally see the expected trends. This helps us conclude that our ideas about the changes in packet traffic are correct, but that there must be other factors affecting the energy usage. I also got some data about the number of data packets being dropped. In this case, there is a strong negative correlation between the number dropped and the SPD. When packets are dropped, they must be retransmitted, and this would require more energy. So this is another factor to look at as a metric for measuring power. Although I did not finally design a model, I was able to narrow some choices down, and make a conclusion about applying the timekeeping metrics directly to a power model. I think this experience has been a valuable one, in more ways than one. First, I have seen firsthand the way graduate students work and learn to analyze and research. Second, I have felt what it is like to be immersed in a completely unfamiliar field of research, and somehow manage my way through it. And finally, by the end, I have an idea of how to start applying my own ideas, draw conclusions and write them up in a moderately intelligible final report!
|
|