"Subconscious Mode" Could Boost Phone Battery Life 85
cylonlover writes "University of Michigan researchers have proposed a new power management system for smartphones that could dramatically improve battery life. The system, known as E-MiLi, or Energy-Minimizing Idle Listening, addresses the energy waste that occurs when 'sleeping' phones are looking for incoming messages and clear communication channels. E-MiLi slows down the clock of a phone's WiFi card by up to 1/16 its normal frequency in order to save power, but then kicks it back up to full speed when information is coming in. The phone uses the header of the incoming message to wake itself up from its 'subconscious mode,' so the clock is at full speed to receive the main message. For users on the busiest networks, it could extend battery life by up to 54 percent."
fractions (Score:4, Insightful)
by 1/16th? or up to 1/16th of? - - quite a difference there!
Re: (Score:1)
Not just fractions.
It doesn't "slow down by up to". My guess is it slows down up to 1/16 of normal clock rate.
The way they wrote it makes it sound like the clock slows down no more than 1/16 of normal frequency. 1/16 clock reduction doesn't sound like it would yield "up to 54%" more battery.
Re: (Score:2)
1/16 clock reduction doesn't sound like it would yield "up to 54%" more battery.
Why not? "Up to 54%" just means less than 54%. So if it provided no improvement whatsoever, that would still be "up to 54%".
I have up to a billion dollars in my pocket.
Old ideas live again (Score:5, Informative)
Patent 4,893,271 [google.com], issued in 1990 and expired, covers an implementation of this idea in which the slow clock is a crystal oscillator, and the high-speed clock is synthesized, using the crystal oscillator as the reference of a phase-locked loop. It was used in tens of millions of Motorola radio pagers for exactly the reason stated in the article -- lower power consumption in sleep mode, while retaining the ability to process fast once a signal appears.
Re: (Score:3)
And even in 1990 the idea that one could slow down the clock of a synchronous digital processor was obvious.
Re:Old ideas live again (Score:5, Informative)
Yes -- the trick is doing it without keeping the high-speed oscillator running all the time. In these kinds of embedded applications, the oscillator can draw a significant amount of power. Implementations that have a high speed oscillator running all the time, but divide it down while in sleep mode, draw more power than the invention, which has a low speed oscillator running all the time, and only generates high frequencies when they are needed.
Re: (Score:1)
which is also painfully obvious to anyone in the field. microcontrollers have had multiple clock inputs and internal oscillator blocks with support for switching between them built in for decades. You can also switch the low speed osc off while the high speed one is running if you're obsessive.
Re: (Score:3)
You can also switch the low speed osc off while the high speed one is running if you're obsessive.
Yes -- if you're willing to pay for two crystal oscillators. The advantage of that patent is that you get by with one crystal.
Re: (Score:1)
Re: (Score:2)
All true. However, while internal oscillators have many good uses, their frequency stability is terrible -- usually something in the range of 0.1%, or 1000 ppm -- so it is difficult to use them when accurate timing is needed. For the application for which this invention was first used, paging receivers, timing accuracy of 100 ppm or so was needed, so a crystal was necessary.
Re:Old ideas live again (Score:4, Interesting)
which is also painfully obvious to anyone in the field. microcontrollers have had multiple clock inputs and internal oscillator blocks with support for switching between them built in for decades. You can also switch the low speed osc off while the high speed one is running if you're obsessive.
Its also painfully obvious to the RF engineering crowd. "obviously" if you have a multiplier chain the high freq mults are going to draw more power than the low freq mults. mults are never known for efficiency so if you have a couple stages you'll find most of your power is dropped in them not the osc. Also you wanna keep the crystal running all the time so its stable, constant temp, etc. Heck keep the first mult stage or so running to keep the load on the osc stable, if you want. That way the calibration output is theoretically usable all the time, instead of being FM modulated as you turn the mults on and off.
Anyway, when you're actively TX or RX, then, and only then, power up the higher freq multipliers in the RX or TX chain (assuming you have 2 separate chains, you just got an instant 50% power savings at the cost of some complicated power wiring). If you're leaving the xtal powered up 24x7 to enhance stability, then powering down the multipliers seems terribly obvious... so that's exactly what they did in ye olden days. When I was a kid I fooled around with some even then obsolete motorola VHF TX and RX strips that had this "obvious" power saving feature.
I don't have a cite, but pick up VHF/UHF/microwave oriented ham radio manuals from about half a century ago, you'll probably be successful.
Merely terminating the last mult stage in a microprocessor clock input instead of a RF mixer is not exactly insightful or patent worthy. Its the equivalent of patenting the concept of using AA batteries in a flashlight instead of their traditional use in kids toys.
The patent is probably much more specific to get around the staggering quantity of prior art. Probably specs the exact layer 2 framing protocol complete with diagrams of the sync header and stuff.
Re: (Score:3)
Oh yeah, if they're so smart why don't RF chipsets include this feature right on the die? What's that you say, they DO? This is is merely the application of an existing technique to a problem solidly within its domain? Don't care, patent approved.
Re: (Score:1)
Gee, I thought that at resonance, very very little power was required to sustain vibration. What am I not understanding about power loss and oscillators, unless the oscillators are mistuned from a natural resonance..
Re:Old ideas live again (Score:4)
Re:Old ideas live again (Score:4, Informative)
I am pretty sure I worked for AT&T on digital pagers that did this well before 1990.
I went on to work on "selective call radios" (cellphones) for Motorola which did the same thing in about 1988.
Quite possible -- the priority date for the patent is 7 November, 1983.
Re:Old ideas live again (Score:4)
This idea still realy sucks balls due to crappy WiFi signals never recieving anything. What we need is an expecting system. If you're not texting a while, just check once per 1,5min and do nothing else in between. Let it have a tickless OS.
Oh wait, we already have that and it is called Linux (tickless)
Should be a realy extremely obvious thing...
Re: (Score:3)
Re:Old ideas live again (Score:4, Informative)
While funny, SMS is being send over GSM, in the same package that is being exchanged with cell towers to maintain GSM (/2G) connectivity, irregardless. That package has empty space, so SMS (that's why it's limited by an amount of characters; so it can fit in that package) can be send without extra load on the cell tower buffer.
What we are talking about here is stuff like Watsapp and Ping and that shit, which is stored on a central server anyway, so when cellphone asks server "What's up, man?" it then gets send without having to fill up the buffer, which is once per 1,5 minutes.
Re: (Score:2)
Re: (Score:1)
It's getting popular with educated people as an ironic word to use. It's also, most definitely a word, it's just an uneducated one. I expect this use was the ironic one, not the uneducated one.
http://oxforddictionaries.com/definition/irregardless [oxforddictionaries.com]
It may surprise you to learn that many of the words we think of as totally acceptable were once like irregardless. Turns out, a few decades of ironic use (or even incorrect use) make it into a new word. I expect decimate will mean devastate shortly, if it
Re: (Score:3)
It may surprise you to learn that many of the words we think of as totally acceptable were once like irregardless.
That doesn't make them any less stupid. Anyone with even a basic understanding of prefixes and suffixes should flag "ir-" + "regard" + "-less" as nonsensical in context. "not" + ("regard" + "without") == "with regard", or worthy of consideration. It's right up there with "I could care less" in the retarded language department.
For the record, I twitch at most uses of the word "decimate" as well. It's rare that one would get to use it properly in most conversation, and rarer still that it would both be us
Re: (Score:2)
Re:Decimate (Score:2)
Re: (Score:1)
Irregardless, educated people often employ improper English ironically.
Re: (Score:1)
Re: (Score:2)
You do realize that due to the lack of an official institute for the English language, there is no standard dictionary and grammer definition, rendering irregardless a perfectly valid English word, because it happens to be used by English speaking people.
Thou art not upset by changing languages, art thou?
Re: (Score:2)
You do realize that words have commonly-accepted meanings, right? Your argument makes as much sense as me going around claiming that up is down and black is white and insist that because there's no institute of official English language my use of those words is perfectly valid.
"ir" == "not"
"regard" == "consideration"
"less" == "without"
ir- + regard + -less == "not (without consideration)" == "worthy of consideration"
Yes, one could argue that "irregardless" is a valid grammatical construction, but it's invar
Re: (Score:2)
Well, strictly speaking; you are correct.
However, since we're not writing articles here and are communication with other preople over the internet; correctness isn't realy required.
Then again, I do not live in an English speaking country, so I'm not getting annoyed about common language rape on a daily basis...
Re: (Score:2)
I don't usually get annoyed about the rape of the language, I'm more annoyed about the willingness of people to make themselves difficult to understand. Using words like irregardless grates, because I have to parse an extra syllable, then figure out they probably meant regardless and discard the whole meaning sense that I was halfway done appending to the sentence!
Grrr.
Re: (Score:2)
Yes, you can use "irregardless" in place of "regardless". You can also use "desalinization" and "preventative" instead of "desalination" and "preventive".
The problem is, you're trying to communicate--to make your message understood by as many readers as possible--and these less-than-preferred forms can cause your readers to pay more attention to the language than to the message.
Not to mention the lack of credibility many readers infer when they perceive an author can't even use the language properly.
Persona
Re: (Score:1)
Re: (Score:2)
While funny, SMS is being send over GSM, in the same package that is being exchanged with cell towers to maintain GSM (/2G) connectivity, irregardless. That package has empty space, so SMS (that's why it's limited by an amount of characters; so it can fit in that package) can be send without extra load on the cell tower buffer.
No.
Allow me to quote a relevant passage from a similar discussion elsewhere: [reddit.com]
Re: (Score:2)
I take it you've never left your phone off for a couple of days while people were texting you, eh?
They already get buffered at the cell company, and all get delivered at the same time when you turn your phone on.
Re: (Score:2)
If you're not texting a while, just check once per 1,5min and do nothing else in between. Let it have a tickless OS.
You do realize that its just ticking on a longer period, right? Its not tickless.
Re: (Score:3)
No, this is something completely different. This is an even slower mode, set to watch only the headers of incoming packets sort of like a Slashdot system, in which people read the first couple of words of an article, then wakes up to comment on said article at full speed to get first post, just with the tiny tweak that it actually reads the rest of the message before replying.
Re: (Score:2)
+5
I try reading 2-3 times before I reply, but this feature still tends to bite me quite often.
WiFi card...? (Score:2)
Re: (Score:2)
TFA repeatedly mentions WiFi.
IIRC, the iPhone only enables wifi at all when you're actively using the phone - when it's sat in your pocket, wifi is off. Which would put a bit of a dent in that 54% figure.
Re: (Score:2)
If you have a decent data plan on your phone, or if you have very low data use (using your phone for e-mail only, for example, will probably not break 50mbyte/month), then there isn't really any point in having wifi on at all. I'll turn it on to update apps, because it's generally faster, but for casual surfing or e-mailing, HSPA (7.2mbit) is fast enough, and uses a fraction of the energy anyway, as most smartphones will already do what TFA is talking about with cellular data: switching to 3G when they need
Re: (Score:2)
This was my thought exactly. But the article REPEATEDLY mentions WiFi, which leads me to believe this research, while maybe novel, is not going to help most smartphone users.
Even if you leave your WiFi on 24x7, it uses a FRACTION of the power of your HSPA/4G coverage. Power use of the 3G radio is the issue, not the wifi radio.
I can leave my phone on wifi all day and if left to idle I will still easily have > 75% battery after 8 hours. With 3G, this would be well below 50%, maybe below 40% if a lot of dat
Re: (Score:2)
Also, it only saves power when you're actively using WiFi. Existing WiFi products already support a power save mode where the receiver is turned off completely for most of the time. It uses a timer to turn itself on briefly to receive beacons, and then turns itself off when there's no traffic indicated. This only leaves the radio on for a few % of the time. Additional power reduction from this mode will be very small.
The new invention is only useful when you're using the WiFi at such an activity level that
Re: (Score:1)
The arithmetic doesn't work out in its favour either. This is essentially a low latency power saving mode. But for 18 hours a day the phone will be sleeping and so should its WiFi. For the 8 active hours you're unlikely to be continuously hammering WiFi and it should be in an existing power saving mode perhaps 6-7 of those hours. The power saving is likely to be 5-10% at
Comment removed (Score:3, Informative)
Re:Just like the owners (Score:4, Funny)
hw r u 2day? batt rnning low lol
Sender is a midget artificial intelligence and is distressed that her bath is leaking.
There's a hole in her bucket, dear Eliza.
Re: (Score:2)
Sender is a midget artificial intelligence and is distressed that her bath is leaking.
Actually, Steve Jobs removed the AI feature from the iPhone, because it used too much battery juice.
What, weren't they doing this already? (Score:3)
Seems like common sense...
Re: (Score:2)
Well, at least in the automotive field that's a standard technique. The article doesn't describe anything particularly new. That might be the problem of the article rather than the problem of the research, though.
Actually to some extent clock-gating is on the way out already: newer manufacturing technologies mean you get increased leakage - i.e. even when you don't switch the gates you draw a high percentage of the current anyway. So instead (or more likely in addition) of controlling the clock you go one
Low Power Listening (Score:2)
These protocols actually turn off the transmitter power and wake it up every t milliseconds to check for activity. When transmitter is turned off for 95% of a second, then you gain 95% battery life when idle
SetCPU in Andoid Market (Score:3)
Re: (Score:2)
CPU clock != Wifi clock
ow about a "not suck" mode (Score:2)
Honestly ALL The phone OS's suck at one thing. Multitasking.
Why cant i white list what multitasks and everything else is forced to close down when I close it? there are very few apps that I want running all the time. No I do not need my shopping list and angry birds running in the background. I want them crashed closed.
There are a total of 5 apps that I would want to ever stay resident and running in the background, everything else I wand dead.
Why cant they do this? it would increase performance and r
Re: (Score:2)
Why cant they do this?
Real OS's as smartphones are only a few years old. Something like systemd on Android could implement something like this (that is, all the parts are available).
But look at the haphazard way Google built Android - they're cleaning up now but time-to-market was the only priority when the started.
This, along with the battery life issue mentioned in TFA and missing encryption, is one of the reasons I still carry a 4-day-battery 'feature' phone.
Re: (Score:1)
This, along with the battery life issue mentioned in TFA and missing encryption, is one of the reasons I still carry a 4-day-battery 'feature' phone.
Android does full-disk encryption (at least 3.0 does - not sure about earlier versiona) and battery life is much improved starting with 3.1 thanks to better CPU throttling and implementation of WiFi power-saving features.
As for the original complaint, there are various apps which can kill tasks either manually or by setting specific criteria. Yeah, it would be nice if the OS had better support for it built-in, and it porbably will eventually, but it's still not an insurmountable problem even now.
Re: (Score:2)
You should try the (2 year old by now) N900 for good multitasking, it is still better in hardware than a lot of newly announced android and w7 phones, and had been optimized and stabilized a lot with the fixes that came all this time.
Would be nice if one of the next community updates includes this method to save battery as an OS patch.
from the horse's mouth (Score:1)
original article: http://www.engin.umich.edu/newscenter/feature/smartphonebattery/ [umich.edu]
research paper: http://kabru.eecs.umich.edu//papers/publications/2011/xyzhang_kgshin_mobicom11.pdf [umich.edu]
New dogs learn old tricks (Score:2)
Well, I guess new dogs can be taught old tricks.
Clocking down a processor to save power while waiting for something to happen is hardly anything new.
I am just surprised that smatrphone developers have just learned how to apply this to what they are doing.
I have an idea (Score:2)
its called the 'off' switch. Maybe I should patent it.
Obligatory BR (Score:2)
Re: (Score:2)
That's not BR, that's PKD. BR never mentions electric sheep.
802.11 (Wifi) has a sleep mode already (FYI) (Score:2)
Wifi clients can have 3 stats:
- active (all normal)
- sleep awake (aka powersave awake)
- sleep asleep (aka powersave asleep)
When the client is in power-save-asleep mode, the access point buffers all the data going to the wifi client.
Meanwhile, the AP sends specific beacons to the client, called TIM.
From time to time, the client goes into powersave-awake mode to get those TIM beacons. It needs a lower power than active mode for the beacons as bandwidth is not a requirement. If the TIM says the client needs t
Wake on LAN? (Score:2)
So this is Wake on LAN, but for WiFi? Wake on WiN? No, theirs was better. Though I'd drop the hyphen: "EMiLi" can more easily be pronounced Emily while "E-MiLi" forces the long-E, giving E-Millie or, worse, if HiFi/WiFi/MiFi carries the day, E-My-Lie (WiDi is already unfortunate: "Why Die").
Title reminds me of the old Little Johnny jokes. (Score:1)
Prior Art (Score:1)
Energy-Minimizing Idle Listening... E-MiLi slows down the clock of a phone's WiFi card by up to 1/16 its normal frequency in order to save power, but then kicks it back up to full speed when information is coming in...
That sounds pretty much like how I make it through Thursdays.
Frequency scaling (Score:2)
This looks to me like CPU frequency scaling, which has been common on desktop PCs since 2004.
Having said that, reducing your CPU by a factor of 16 does seem much more impressive than the 2-3 which is usually employed.