TailTheft

Abstract

A notion in cellular communications (e.g., GSM, 3G) is called Tail Time, namely the period of high power state after the completion of a transmission. This Tail Time can alleviate the overhead of switching from the low to the high power state, in case there is another transmission in the near future. However, it also results in large energy waste. In this paper, we propose TailTheft, a scheme that steals the wasted Tail Time for data prefetching and delayed transfer, thus significantly reducing energy consumption. In order to achieve this, TailTheft schedules a number of transmissions to the Tail Time of other transmissions. A Virtual Tail mechanism is employed to steal the Tail Time and a Dual Queue Scheduling algorithm is proposed for the scheduling of transmissions. We evaluate TailTheft through a trace-driven simulation using both per-application and real life traces. Experimental results show that TailTheft can reduce energy consumption by 20% to 34% in different cases.