Energy-aware cross-layer burst buffering for wireless communication

Abstract

The massive explosion of mobile applications with the ensuing data exchange over the cellular infrastructure is not only a blessing to the mobile user but also has a price in terms of regular discharging of the device battery. A big contributor to this energy consumption is the power hungry wireless network interface. We leverage a measurement kit to perform accurate physical energy consumption measurements in a third generation (3G) telecommunication modem thus isolating the energy footprint of data transfers as opposed to other mobile phone-based measurement studies. Using the measurement kit we show how the statically configured network parameters, i.e., channel switch timers, and buffer thresholds, in addition to the transfer data pattern and the radio coverage, impact the communication energy footprint. We then demonstrate that being aware of static network parameters creates room for energy savings. This is done by devising a set of algorithms that (a) infer the network parameters efficiently, and (b) use the parameters in a new packet scheduler in the device. The combined regime is shown to transfer background uplink data, from real world traces of Facebook and Skype, with significant energy saving compared to the state-of-the-art.

When mice consume like elephants

“…Inactivity timers create energy overheads known as energy tails since the UE remains in a high energy consuming state while not transmitting anything [6,25]. Fig.…”

“…These shift transmissions over time (e.g., batching background traffic) to minimise the transmission cost. For example, our previous work [25] schedules background data transfers considering the inactivity timers and the RLC data buffers in combination.…”

Kernel level energy-efficient 3G background traffic shaper for android smartphones

“…These transmissions are aggregated and sent in a single burst. In particular, the Cross-Layer Burst Buffering (CLBB) algorithm [1] schedules background uplink traffic in an energy-efficient manner using this general idea. Given a maximum waiting time for a packet transmission (T w ), CLBB uses the knowledge of inactivity timers and RLC buffer thresholds to perform data aggregation: when the UE is in PCH, CLBB does not send packets until their T w is about to expire, maximising the time in the lowest consuming state (PCH) and reducing the number of energy tails.…”

“…There exist several approaches [1]- [8] to potentially reduce the energy consumption of these cellular data transfers that do not have any quality of service requirements. Most of these rely on shaping the traffic.…”

“…The contributions of our work are: (a) realization of an existing algorithm (from [1]) on the Android platform, (b) measuring the energy footprint of the algorithm operation, and (c) discovery of the implications of using such a solution with real traffic generated during applications' operation (which cannot be steered towards the sort of traffic that it was designed for), thereby identifying future directions of research.…”