Researchers at the University of Michigan have developed a new smart phone mode that could extend battery life by more than 50 percent. I spoke this week with Kang Shin, a computer science and engineering professor, who presented a paper on the work yesterday at the ACM International Conference on Mobile Computing and Networking in Las Vegas. Below are excerpts from our interview.
On why smart phones use battery power, even when we're not using them:
Phones are not just phones anymore. They do email, web browsing, music and media. The advances in battery technology have been lagging significantly behind the advances in hardware technology. I have an iPhone for my personal use. My iPhone battery doesn't last a whole day.
When I don't send or receive messages, what does my phone do? The phone doesn't know when a message will be arriving, but I want to receive the message as soon as it comes. My phone listens to all the incoming messages through the air. Idle listening means I'm not receiving any messages, but I have to listen to see whether a message is for me. Idle listening uses almost as much power as active transmissions. That's a huge issue. Even if I don't use my iPhone during the day because of a meeting or class, it's continuously listening to messages to find out if one is coming to me.
On why sleep mode isn't the answer:
Sleep means you're putting hardware devices into subconscious mode, running at very low frequency. If [a message] comes when you're in sleep mode, you won't notice. If you stay sleeping, you won't be able to receive anything. You could sleep for 10 minutes, wake up to check for messages and go back to sleep again. You can set these intervals for yourself. For the time you're sleeping, you're going to consume less energy. But even with sleep mode, the battery might run out halfway through the day.
On reducing power consumption during idle listening:
Eighty percent of cell phone users keep their cell phones in idle listening mode for 90 percent of the time. We looked at ways to reduce power consumption. Can we somehow put the phone in subconscious mode, but still ensure it can identify messages sent for you? In other words, even if you're in subconscious mode, you'll hear it if someone calls you. Once you hear it, you're supposed to fully wake up. But if you're in subconscious mode, you're not supposed to hear or understand anything. We decided to operate these devices at a lower frequency. If you do this, the power use of the device will be reduced proportional to its operating frequency. The problem is: If your name is called, you won't be able to comprehend it.
We wanted to separate detection of my identity from the entire message I receive. I want to put my phone in idle listening mode with the minimum capability to comprehend the identity, but not the message. Once I identify my name in an incoming message, I'll fully wake up. The sender should have a preamble or header that contains your identity and other key information. Right now, the identity is encoded in the preamble and you won't be able to encode it unless you are operating at full frequency. We created a small preamble using the device identity. The preamble is detectable by the receiver even in subconscious mode. This is energy-minimizing idle listening. Battery life can be extended by 54 percent.
On the next steps for this work:
We've implemented this already. Every phone has a receiving part and a transmitting part. The receiving part is easy to do. Most of the central processing units now come with the capability of changing the operating frequency.
When I want to send you a message, I have to include that preamble. We have to modify the firmware of the sending side of the phone. We don't have chip manufacturing capabilities at the University of Michigan. We can't change the firmware of existing phones. We used a device and implemented our algorithm of generating this preamble. We have to polish up that part of the software code and run that code into the smart phone chip.
We've applied for U.S. and international patents. I've gotten many requests from companies who want to license it. Some venture capitalists want to invest in it. I'm going to think about it.
Photo: Kang Shin