Boffins boost battery life with underclocked Wi-Fi
Removing the idle listening energy tax
Scientists from the University of Michigan have devised a power management system that can greatly improve the lifespan of radio devices such as smartphones and laptops.
Engineering professor Kang Shin and doctoral student Xinyu Zhang have developed a technique for saving the battery life of Wi-Fi devices by clocking down the Wi-Fi radio system to 1/16th of its normal idling power output.
Dubbed "Energy-Minimizing Idle Listening" (E-MiLi), the technology works with current Wi-Fi technology (when the correct software is installed), and it can extend the life of a battery by up to 54 per cent, according to the boffins.
“We have a proof-of-concept implementation of E-MiLi based on software radios,” Xinyu Zhang tells The Register in an email exchange. In testing, the technique can boost smartphone battery life by 54 per cent.
Even in idle mode, standard Wi-Fi systems use more than two-thirds of the power consumed when fully operational, because the radio needs to be able to accept new data by examining incoming messages. The researchers call this "idle listening". E-MiLi allows the radio to largely power down, but it can reactivate by investigating header information on new data rather than examining all the data as a lump.
Specific tags in the header are needed and to make it work in existing kit, firmware and device drivers must be updated for the Wi-Fi card embedded in those mobile devices. The Wi-Fi cards also need to support downclocking at the hardware level.
"We came up with a clever idea," Shin says. "Usually, messages come with a header, and we thought the phone could be enabled to detect this, as you can recognize that someone is calling your name even if you're 90 per cent asleep."
Shin and Zhang present their paper on September 21 at the ACM International Conference on Mobile Computing and Networking in Las Vegas. They are now working on other power-saving techniques for Wi-Fi, specifically looking at regulating voltage directly.
“We have not completed experiments with voltage scaling. We expect much higher gain, because the power consumption is proportional to clock rate, and proportional to the square of voltage,” Xinyu Zhang says. ®
Sponsored: Hyper-scale data management