From Supercapacitors to Lithium-ion Capacitors: State-transition-aware energy management in energy harvesting systems

Abstract

Lithium-ion Capacitors (LiCs) have emerged as promising energy storage devices for energy harvesting applications due to their high power capability, extended cycle life, and higher energy density compared with conventional Supercapacitors (SCs). Existing energy management strategies developed for SC-based systems, such as the Modified Earliest Deadline First (MEDF) algorithm, exploit charge redistribution phenomena to improve task scheduling by adapting execution decisions according to the internal energy state of the storage device. However, these approaches rely on assumptions derived from the behavior of SCs, where voltage relaxation is mainly governed by capacitive charge redistribution. Due to the hybrid capacitive and battery-like mechanisms of LiCs, these assumptions may no longer accurately represent the evolution of the available energy. This paper investigates the applicability of charge redistribution-based scheduling strategies to LiC-based energy harvesting systems and demonstrates the limitations of directly extending MEDF rules to LiCs. An experimentally parameterized LiC model is used to analyze the transient voltage response and energy availability during different operating states. Based on this analysis, a State-Transition-Aware Earliest Deadline First (STA-EDF) algorithm is proposed, where task execution decisions are determined according to the predicted evolution of the LiC state rather than predefined redistribution rules. The proposed approach is evaluated against conventional EDF and MEDF strategies under different energy harvesting and load scenarios. Results show that, while STA-EDF and MEDF achieve comparable energy violation rates on the investigated LiC, STA-EDF replaces a redistribution-based decision criterion that vanishes below the scheduling timescale with a physically grounded trajectory-based criterion. The work further establishes that per-task scheduling benefit on a LiC is fundamentally bounded by its limited state recovery during short harvesting intervals, motivating trajectory-based and energy-budgeting approaches over redistribution-based ones.