Abstract
While cellular interfaces are known to be the main power consumers in smartphones, our measurements reveal that surprisingly, due to cellular tail effect, a substantial ratio of energy drain comes from the low-frequency, low-data-rate background traffic. To reduce cellular tail energy use, existing solutions either require changing application behavior or assume the availability of low-level control of cellular interface (e.g., fast dormancy). We propose a collaborative cellular tail energy reduction approach for background traffic, which opportunistically discovers neighbors using low-power Bluetooth radio and shares their cellular bandwidth. Furthermore, we define a rigorous fairness notion by adapting the generalized processor sharing concept. Our sharing scheduling algorithm uses this notion to achieve substantial energy saving while ensuring fairness for participating phones and is also robust to whitewash attacks. Our evaluation demonstrates up to 90% saving for background traffic energy consumption in urban settings.