SMARTPHONE BATTERY POWER PREDICTION BASED ON CONTINUOUS-TIME AND USAGE PATTERN MODELS
Keywords:
Smartphone battery, Continuous-time model, Usage pattern, Thermal-energy coupling, Power consumption predictionAbstract
This paper develops a smartphone battery power prediction framework based on the continuous-time model and the usage pattern model. The continuous-time model decomposes useful power into screen, CPU, network, GPS, and RAM components, then incorporates temperature-dependent Joule heating, active thermal management, and self-discharge into a coupled thermal-energy differential system. This structure describes how hardware settings, usage behavior, ambient conditions, and temperature jointly determine real-time power drain and battery energy decay. The usage pattern model further characterizes seven typical scenarios, including idling, gaming, web browsing, navigation, communication, multimedia playback, and camera operation. Based on a public dataset covering the daily feature usage of Android devices, model parameters are divided into usage pattern parameters, static hardware parameters, and power-type hardware parameters. The Trust Region Reflection Algorithm is used to invert hardware coefficients that cannot be directly obtained, while the Analytic Hierarchy Process evaluates the relative intensity of ten core variable parameters across the seven usage patterns. The analysis shows that screen power dominates most screen-on modes except Idle, making display power reduction a priority for power saving. Thermal management alleviates overheating but shortens battery life; in Navigation mode, it reduces the maximum temperature from about 43°C to 39°C while decreasing battery life by about half an hour.References
[1] Flores-Martin D, Laso S, Herrera J L. Enhancing Smartphone Battery Life: A Deep Learning Model Based on User-Specific Application and Network Behavior. Electronics, 2024, 13(24): 4897.
[2] Neto A S B, Farias F, Mialaret M A T, et al. Building Energy Consumption Models Based on Smartphone User's Usage Patterns. Knowledge-Based Systems, 2021, 220: 106908.
[3] Dash P, Hu Y C. How Much Battery Does Dark Mode Save? An Accurate OLED Display Power Profiler for Modern Smartphones. Proceedings of the 19th Annual International Conference on Mobile Systems, Applications, and Services, 2021: 323-335.
[4] Guegain E, Raes R, Chachignot N, et al. AndroWatts: Unpacking the Power Consumption of Mobile Device's Components. 2025 IEEE/ACM 12th International Conference on Mobile Software Engineering and Systems (MOBILESoft), 2025: 71-81.
[5] Gopal K R, Ma B, Bai P. Mapping Out Fast Charging Safe Limits for High-Loading Lithium-Ion Cells by High-Fidelity Operando Microscopy. Small (Weinheim an der Bergstrasse, Germany), 2026: e14619.
[6] Liu S, Li Y, Li S. Effective anode protection for zinc-ion batteries using a covalent–organic framework coating strategy. Electrochimica Acta, 2026: 148191.
[7] Radhakrishnan A, Prasad V R, Railis J D, et al. Enhancing thermal management in electric vehicles to improve performance and extend battery life using advanced cooling systems. Journal of Thermal Analysis and Calorimetry, 2026(prepublish): 1-20.
[8] Zhang Y, Sun Z, Cai C. Real-time adaptive hybrid energy management for hybrid-electric UAVs: Offline–online DP-ECMS collaborative strategy in low-altitude inspection. Franklin Open, 2026: 100473.
[9] Samal B K, Pati S, Sharma R. Power management using an improved EMS algorithm in a stand-alone hybrid PV-PEMFC microgrid with reduced converter count. Green Energy and Intelligent Transportation, 2026, 5(2): 100302.
[10] Huang K, Zhang X, Guo Y, et al. Source domain selection with early-cycle features for transfer learning-based prediction of lithium-ion battery degradation trajectories. Energy, 2026: 139928.
[11] Monteiro E, Souza E, José R, et al. A dataset from the daily use of features in Android devices. Mendeley Data, V6, 2024. DOI: 10.17632/bpsrw76hgx.6.