Aggregating inter-app traffic to optimize cellular radio energy consumption on smartphones

Abstract

Cellular radio interfaces on smartphones consume a significant amount of battery power, specially with growing number of network centric applications. With high bandwidth cellular access links pushing the bottleneck to the network core, the risk of poor bandwidth utilization of the access link increases, which leads to energy wastage as the radio interface must stay active longer. In addition, small sized packet transmissions from applications wake up the interface frequently, but do not require the entire bandwidth capacity. In this work, we improve the radio usage by aggregating packet transmission from multiple applications. We introduce different time delays while transmitting packets from foreground and background applications such that user experience is minimally impacted. Through empirical observations, we determine the impact of different types of traffic on bandwidth utilization. Naive attempts to improve bandwidth utilization lead to increase in the number of packets missing the deadline for dispatch. With these observations, we propose a technique that balances the bandwidth utilization and deadline misses. Simulation driven experiments using synthetic traces and real trace based on application usage on Android based smartphones show energy gain of around 10% over other competing techniques.