Researchers at the University of Michigan have come up with a way to extend the battery life of tablets, smartphones and other devices that use WiFi.
Kang Shin, a professor of computer science and engineering, and Xinyu Zhang, a doctoral student, have developed E-MiLi, a power management method that could cut energy consumption by about 44 percent for up to 92 percent of users in WiFi zones.
E-MiLi, or Energy-Minimizing Idle Listening, involves slowing down the rate at which the WiFi receiver retrieves packets, along with filtering out unnecessary packets.
That filtering is done by creating special headers that include the destination address of each packet, and then applying an algorithm developed by Shin and Zhang that detects packets addressed specifically to a particular receiver and wakes up the receiver only then.
“It’s a clever little discovery,” Carl Howe, a director of research at the Yankee Group, told TechNewsWorld.
E-MiLi works for “all WiFi-equipped mobile devices, including mobile phones, tablets and laptops, and can also be used by WiFi networks such as ZigBee,” said Zhang.
“It significantly improves the energy efficiency of WiFi devices,” he told TechNewsWorld. “For example, it extends battery life by 54 percent for smartphones.”
The E-MiLi Art of Selective Snoozing
Since it’s not possible to predict when packets will arrive at a WiFi receiver, Shin and Zhang decided to reduce the clock rate of the receiver during its IL period and have it wake up and respond to incoming packets as they arrive.
However, receiving packets at a lower clock rate is a problem, because the Nyquist rate requires that the receiver’s sampling clock rate must be at least twice the bandwidth of the transmitted signal.
So, E-MiLi uses a new approach called “Sampling Rate Invariant Detection,” or SRID.
This adds a special preamble or header to each packet of data and incorporates a linear-time algorithm that can accurately detect the preamble even if the receiver’s clock rate is much lower than that of the transmitter.
SRID embeds the destination address into the preamble so that a receiver will only respond to packets destined for it.
On detecting the preamble, the WiFi receiver rockets into full clock rate and recovers the data packet.
“Everybody forgets WiFi is based on Ethernet, and there never was any attention paid to this preamble and power-saving stuff because it was designed for a world of desktop and server computers,” the Yankee Group’s Howe pointed out.
“When they made it wireless, they probably should have spent more time on this,” he added.
Why We Need E-MiLi or Something Similar
WiFi’s power-saving mode doesn’t help much, because it can’t reduce the idle listening (IL) time associated with carrier sensing and configuration, Shin and Zhang found.
IL consumes as much energy as active transmission and reception, and WiFi clients spend lots of time in IL because of technical issues such as media access control (MAC)-level contention and network-level delays, noted Shin and Zhang.
Even with PSM enabled, IL accounts for more than 80 percent of energy consumption for clients in a busy network and 60 percent in a relatively idle network.
WiFi receivers must constantly be in IL mode because packets arrive unpredictably, and also because WiFi receivers must find a clear receiving channel.
That’s true for all wireless radio receivers — for example, Bluetooth hops among 79 channels in order to minimize receiving errors, according to a dissertation by Texas A&M University student Ahmed Ahmed Emira. Each Bluetooth packet has to arrive within a 625-microsecond slot.
Where E-MiLi Might Go
E-MiLi lets SRID be integrated into existing MAC or sleep-scheduling protocols by adding a downclocked IL mode into the receiver’s state machine through Opportunistic Downclocking (ODoc).
ODoc takes a smart approach to downclocking, assessing the potential benefit of doing so before letting the receiver downclock.
Shin and Zhang’s tests show that E-MiLi can detect packets with close to 100 percent accuracy even if the receiver operates at one-sixteenth the normal clock rate.
E-MiLi reduces energy consumption consistently across different traffic patterns without any noticeable performance degradation, Shin and Zhang found.
Device manufacturers will likely show the most interest in E-MiLi, said Zhang, because it runs in the hardware, firmware and device drivers of WiFi cards.