ESS: Repeatable Evaluation of Energy Harvesting Subsystems for Industry-Grade IoT Platforms

Designing energy-harvesting Internet of Things (IoT) systems is challenging because application performance depends crucially on the amount of energy supplied to the system. Reliably comparing application and hardware design options hence requires controlling the energy supply. IoT devices are however becoming increasingly complex and typically feature highly optimized and proprietary communication protocol implementations. Existing evaluation approaches either rely on hardware prototypes — which support industry-grade protocols but exploring energy supply subsystem design choices requires building a prototype for each option — or simulators — which enable exploring all aspects of the design but cannot support proprietary protocols. We hence propose the Energy Harvesting Subsystem Simulator (ESS) which bridges the gap between these approaches by simulating the energy supply subsystem in a trace-based manner while measuring the power consumption of an industry-grade SoC — thereby enabling repeatable evaluation of energy harvesting subsystem design options for industry-grade IoT platforms. We further propose the Compress Inactive (CIN) optimization for periodic IoT applications which speeds up ESS evaluation time by up to 516.0× compared to real-time evaluation while retaining an average time (energy) prediction error of only 0.2% (0.8%).