SwitchR: Reducing system power consumption in a multi-client, multi-radio environment

Abstract

Multiple wireless network interfaces in a single mobile device exist in order to support their diverse communications and networking needs. This paper proposes a general switching architecture, SwitchR, for managing radio communications for multiple (client) devices utilizing multiple heterogeneous radios per device. To be effective and useful, the radio switching must be transparent to applicaticons, and should be deployable incrementally within existing wireless infrastructures. The SwitchR framework considers the load imposed on the wireless channel by other communicating clients in order to make optimal switching decisions. We demonstrate reduction in energy consumption of a mobile device by 47% -72%, depending upon the application, over the Power Save Mode in WiFi. SwitchR also leads to 13% -60% reduction in energy consumption over previous multi-radio architectures that do not consider the interactions between multiple clients.

Bluesaver: A Multi-PHY Approach to Smartphone Energy Savings

“…Friedman et al [28][29] provide a study on power and throughput tradeoffs of WiFi, Bluetooth, and other interfaces in smartphones, but do not go beyond offering theoretical analysis and insights for phone developers. SwitchR [30] devises a multi-client switching policy enabling communication among clients within a multi-radio environment. However, the work does not go far beyond providing the framework and policies.…”

eBond

“…Wireless network energy has been studied extensively in the setting of mobile devices [1]. The resulting models, which weigh throughput and power consumption with other factors prevalent in mobile settings such as battery lifetime, tend to become very complex [5].…”

“…Not only is bonding a technique that is available in most data center switch technology but it also allows us to not care about modifications of the routing tables or similar intricacies of the network stack, that appear when switching between network interfaces [1].…”

Energy Efficiency of Connected Mobile Platforms in Presence of Background Traffic

“…Prior work can largely be grouped in three categories: reducing the active power consumption of systems when they are awake Agarwal et al (2008); Flinn & Satyanarayanan (2004); Li et al (2007), reducing the power consumption of the network infrastructure, routers and switches Gunaratne et al (2005); Gupta & Singh (2003); Nedevschi et al (2008) and opportunistically putting the devices to sleep Agarawal et al (2009);Shih et al (2002); Sorber et al (2005). Long Idle falls in the third category, where it advocates for localized energy-efficient optimization within the platform to extend the sleeping state of the platform rather than using a network-wide implementation of a proxy (wakeup) service.…”