TY - GEN
T1 - Operation policies for energy harvesting devices with imperfect state-of-charge knowledge
AU - Michelusi, Nicolo
AU - Stamatiou, Kostas
AU - Badia, Leonardo
AU - Zorzi, Michele
PY - 2012
Y1 - 2012
N2 - As Energy Harvesting Devices (EHD) become more widely deployed in sensor network platforms, the need arises for 'smart' operation policies which can ensure long-term, autonomous and reliable operation. Existing research has relied on the implicit assumption of perfect knowledge of the energy available in the EHD. However, estimating the energy level of the batteries or super-capacitors employed in real-world EHDs, commonly known as State-Of-Charge (SOC), is a non-trivial task. In this paper, we design operation policies that maximize the long-term reward under imperfect knowledge of the SOC. Through an array of simulation results, we quantify the performance degradation due to imperfect SOC knowledge, and show that it increases with decreasing storage capacity and increasing variance in the energy arrival process. In the particular case of a two-state controller, i.e., a controller which knows only if the SOC is HIGH or LOW, we prove that, for a linear reward function, there is no performance loss, while, for a logarithmic reward function, simulations show that the loss is typically less than 5%.
AB - As Energy Harvesting Devices (EHD) become more widely deployed in sensor network platforms, the need arises for 'smart' operation policies which can ensure long-term, autonomous and reliable operation. Existing research has relied on the implicit assumption of perfect knowledge of the energy available in the EHD. However, estimating the energy level of the batteries or super-capacitors employed in real-world EHDs, commonly known as State-Of-Charge (SOC), is a non-trivial task. In this paper, we design operation policies that maximize the long-term reward under imperfect knowledge of the SOC. Through an array of simulation results, we quantify the performance degradation due to imperfect SOC knowledge, and show that it increases with decreasing storage capacity and increasing variance in the energy arrival process. In the particular case of a two-state controller, i.e., a controller which knows only if the SOC is HIGH or LOW, we prove that, for a linear reward function, there is no performance loss, while, for a logarithmic reward function, simulations show that the loss is typically less than 5%.
UR - http://www.scopus.com/inward/record.url?scp=84871943610&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84871943610&partnerID=8YFLogxK
U2 - 10.1109/ICC.2012.6364958
DO - 10.1109/ICC.2012.6364958
M3 - Conference contribution
AN - SCOPUS:84871943610
SN - 9781457720529
T3 - IEEE International Conference on Communications
SP - 5782
EP - 5787
BT - 2012 IEEE International Conference on Communications, ICC 2012
T2 - 2012 IEEE International Conference on Communications, ICC 2012
Y2 - 10 June 2012 through 15 June 2012
ER -