PREDICTION AND ASSESSMENT OF AMBIENT ENERGY SIGNALS FOR ENERGY HARVESTING SYSTEMS

Open Access
Author:
Mohapatra, Jagruti
Graduate Program:
Computer Science and Engineering
Degree:
Master of Science
Document Type:
Master Thesis
Date of Defense:
July 07, 2016
Committee Members:
  • Vijaykrishnan Narayanan, Thesis Advisor
  • John Morgan Sampson, Committee Member
  • Vijaykrishnan Narayanan, Committee Member
Keywords:
  • Prediction
  • energy harvesting
  • Non-volatile processors
  • backup
Abstract:
Prediction of Wi-Fi and cellular signals is a challenge as the signals fluctuate with time. There is no definite trend or seasonality associated with the signals. The prediction of random stochastic signals has always been a challenge. Wi-Fi and cellular signals are ambient energy sources and can be used for energy harvesting non-volatile processors. This thesis discusses the prediction of such stochastic events using statistical prediction algorithms to predict the outcome of next few time periods. Based on these predicted power profiles, I am calculating the energy required to power an NVP module and check when the processor backs up, restores or runs. Hence, it becomes imperative on our part to predict these signals such that there is a healthy balance between performance and overheads. I created a simple NVP simulator to calculate the number of backups and the number of restores required for both the actual and the predicted energy profiles. A comparison was made and was found that the model could almost correctly predict the backup cycles.