Performance Bounds for Remote Estimation under Energy Harvesting Constraints

Ayca Ozcelikkale, Chalmers

Abstract:  We consider a remote estimation problem under energy harvesting (EH) constraints. The sensor node observes an unknown field and communicates its observations to a remote fusion center using uncoded transmission. Due to unreliable nature of the energy available, the sensor has to find the optimal trade-off between using the available energy for its current operations and saving its energy for the future. The distortion minimization problem at the fusion center is considered. We focus on the mean-square error as the performance criterion. Contrary to the traditional approaches, the degree of correlation between the signal values constitutes an important aspect of our formulation. We provide the optimal sensor power allocation strategies for a number of illustrative scenarios and low-complexity schemes for others. We provide performance bounds on the achievable distortion under a slotted block transmission scheme, where at each transmission time slot, the data and the energy buffer are completely emptied. Our bounds provide insights to the trade-offs between the buffer sizes, the statistical properties of the energy harvesting process and the achievable distortion.  In particular, these trade-offs illustrate the insensitivity of the performance to the buffer sizes for signals with low degree of freedom and  suggest performance improvements with increasing  buffer size for signals with relatively higher degree of freedom.  Depending only on the mean, variance and finite support of the energy arrival process, these results provide practical insights for the battery and buffer sizes for deployment in future energy harvesting wireless sensing systems.


Ayca Ozcelikkale is currently a post-doctoral researcher at Department of Electrical Engineering at Chalmers University.  She has received her Ph.D. degree from Electrical Engineering, Bilkent University, Turkey. During August 2010 - June 2011, she has visited Mathematics and Statistics Department, Queen's University, Canada. Her research interests are in the areas of communications, statistical signal processing, compressive sensing and optimization. In 2015, she has been awarded Junior Project Research Grant by Swedish Research Council (VR) with the project "Smart Sensing and Communications with Energy Harvesting'' for 2016-2019. She also had a recently concluded EU Marie Skłodowska-Curie Fellowship with the project title "GRENHAS: Green and Smart Communications with Energy Harvesting: A Signal Processing Approach”.