Protection-Aware Electro-Thermal Modeling for Smartphone Battery Time-to-Empty Prediction

Abstract

: Accurate smartphone time-to-empty (TTE) prediction cannot be obtained by extrapolating state of charge (SOC) alone. Near low SOC, under high load, at low temperature, or after battery aging, terminal-voltage sag and increased internal resistance may make demanded power infeasible before stored charge is depleted. This paper presents a compact protection-aware electro-thermal framework for TTE prediction from smartphone usage logs. The framework follows one engineering chain: usage logs are reconstructed as a continuous demanded-power trajectory, demanded power is closed through an electro-thermal battery model, the deliverability margin drives runtime throttling and undervoltage protection, and TTE is computed as the first infeasible operating time. The model addresses two research questions: how usage logs can be transformed into P dem (t), and how deliverability constraints and operating-system protection improve realistic TTE prediction. The formulation combines interpretable demand reconstruction, OCV/internal-resistance power closure, SOC and thermal dynamics, aging-dependent capacity and resistance, soft power clipping, and UVLO-based shutdown. It explains non-zero SOC shutdown and provides a physics-constrained basis for evaluating runtime under different loads, temperatures, and aging states.