Green Mobile Computing: A Review of Energy-Aware Smartphones, Cellular Radio Access, Edge Intelligence and Smart Grid Analytics

DOI

https://www.doi.org/10.59256/ijsreat.20260606019

Published In

Volume 6, Issue 6 (2026)

Published Online

November-December 2026

Pages

131-138

Article Type

Review Article

Check for updates

Mobile devices, cellular networks and the computing services behind them now form a large and growing source of electricity demand in the information and communication sector. A smartphone drains its battery through the screen, the radio and background applications, while the network that serves it spends most of its energy in base stations that stay powered even when traffic is light. Research on these problems is split across mobile software, wireless networking, edge computing and power-system communities, each optimizing its own layer. The objective of this review is to bring these strands into one picture that follows energy from the handset, through the radio access network and the edge or cloud, to the buildings and grids that supply the power. The review organizes 41 studies into a five-layer taxonomy and compares their aims, methods and reported outcomes. Device-level studies point to screen state, radio signalling, application wake-ups and user behaviour as the main drivers of handset energy use, and a mapping study found user behaviour to be the factor most often linked to battery drain. Network studies show that sleep modes and lean carrier designs can cut a large share of radio access energy, and that one set of architectural measures could reduce total mobile network energy by roughly a third. Edge studies report that energy-aware offloading and federated training can reduce system energy, with one study reporting savings of up to 59.5%. Grid studies show that machine learning now supports load forecasting, smart metering and the control of hybrid renewable plants. Few studies measure energy across more than one layer, and the energy cost of the learning models themselves is rarely reported. Future work should develop cross-layer energy accounting, shared benchmarks and lightweight models that remain efficient on constrained devices.