Slashdot Mirror
Your Computer and Cell Phone Are Lying To You
Ant writes with a story from Dan's Data, which says that the battery meter and connection-strength displays in your portable electronics are lying to you, "and not just when they whisper to you in the night." Quoting: "Mobile phones, and most modern laptops, have signal strength and battery life displays. One or both of these displays has probably been the focus of all of your attention at one time or another. Neither display is actually telling you what you think it's telling you. The signal strength bars on a mobile phone or laptop do, at least, say something about how strong the local signal is. But they don't tell you the ratio between that signal and the inevitable, and often very considerable, noise that accompanies it ..."
And I bet you're going to tell us next that DRM isn't for our own good and is just a way for conglomerates to steal more of our money with little effort done on their part. Hah!
The article was indeed interesting, and believable. But it has a bad case of [Citation-Needed].
Belief? Hope? Preference?The Existential Vortex
Who cares? When it's full, my laptop or cellphone works great. When it's empty, the thing stops working. When there's only a few bars left, I either plug it in / move to a different location. IMO, it perfectly performs its intended duty. Anything beyond that is geek pedantry and nitpicking.
An old-timer with old-timey ideas.
And I even have a little meter for it mixed in with my signal strength.
I find it pretty useful.. I'm pretty sure everyone's wireless chipset can tell them how much noise or at least how many mangled packets arrive. It's just the little dummy strength meter doesn't convey any of that. I liken most of those sorts of things to the CEL light in cars anyway. Good to know when something's not *perfect* but not so good for understanding why (nor whether it's just a gas tank cap seal broken, or a head gasket blown.)
Exsqueeze me?
I've written a wifi signal strength meter for an embedded product. During my research, I found it was pretty much standard to base the bumber of bars on the signal to noise ratio, not the raw signal strength.
This article gives me a hunch why my no-name laptop battery dies so quickly even when Ubuntu still thinks it has 10% charge and several minutes left. Didn't happen with the manufacturer's battery...
Ubuntu usually does an excellent job analysing how good your battery really is (not sure if it's the kernel ACPI or HAL or GNOME that's actually doing it). But when the battery lies so blatantly, it seems even Ubuntu can't keep my laptop from sudden death without a proper warning or shutdown.
Both my Blackberry and my Sony Ericsson sometimes decide not to connect a call when I have close to full signal. Judging from TFA this could then be because of high noise ratio.
At the same time, I have always wondered why my phones do not give me any indication why the calls were not connected at the time. They both just return to the main screen after a long period of connection attempts.
The engineers dilemma, at least for battery levels:
- how the real value taking into account all variances including current usage and thus constantly move up and down the value
- average out the results to something close, but not exact, since this is what satisfies most people
Jumpstart the tartan drive.
Is it just my luck or are all cars like that? You go 200km on the first 25% of the gauge, but can barely get to 550km before it's empty?
Dan doesn't seem to know much about batteries. Check out batter power discharge curves and such...
http://www.mpoweruk.com/performance.htm Remaining power is estimated based on the charge of the battery. If you notice on those graphs, when you get out to the end of the stored charge, it drops off very quickly, which is why the gauge goes from half to empty quickly.
It really does not matter what these meters say as long as they are consistent. From long experience, my grey-ware then interprets the bars to give me a realistic expectation of battery life or signal strength. Move along now please. Nothing of interest here.
Are you trying to tell me that the constantly changing field of electro-magnetic radiation pouring through my laptop does not always match up precisely to the five bars in the display? Frankly, I find that hard to believe.
Tic-Tac-Toe, Global Thermonuclear War, and relationships all have the same winning move.
Comment removed based on user account deletion
I would have given you the answer, but I suppose you would rather that I speak axiomatic set theory or such.
Signal strength and battery time remaining can get pretty complicated, the more you look into it. There are a ton of different measurements, historical information, performance expectations, etc. that are constantly changing based on how the device is being used, who is using it, etc. At some point, you need to condense all of that information into some pretty little bars that a *normal* user (i.e. someone who has never heard of Slashdot) can comprehend. Is there going to be some precision lost? Of course. Is the graphical representation going to convey all the data gathered and interpreted by the device? Of course not. But the idea is to make it as useful as possible.
Just because I can hook a shark from a boat, I do no offer to wrestle it in the water.
Cingular loves to tout "More bars in more places".
"Higher signal-to-noise ratio across a broader range of the United States" just isn't quite as catchy a slogan.
#1, even with a voltmeter you can't reliably predict battery life. With an alkaline AA battery, you could watch the voltage drop from 1.5V down to 1.1 and know that it was now dead - but with newer rechargeable batteries, the voltage doesn't drop until it's completely dead, so you can't easily guess how long it will take. The only way to do it would be to have the device keep a history of how long it is able to work before the battery dies completely and statistically predict future performance. As if they are going to waste time doing that!
#2 Yes, noise should be considered, but an exact signal to noise ratio isn't going to predict bandwidth or call quality, either. I'm pretty sure that the "signal" they measure is actually signal-to-noise anyway. But even just signal strength is still useful, since you can assume that noise isn't changing that much.
Gas gauges? How many people see that their car stays "full" for a long time and then drops sharply? Or says that it is empty when there's still a few gallons left? Mine will tell me "0 miles to empty" and drive for another 50 miles without coming close to empty. Speedometers? They can be off by 5 or 10% right from the factory. Really every gauge is inaccurate by some amount.
My guess is that companies make the gauges vague on purpose, so that people DON'T try to get too much (false/misleading) information out of them. If your cell phone can make a phone call with "2 bars" of signal, that is all the information you should be taking away from that measure. And if your battery says full for 2 days and drops sharply on day 3, you know that when it starts to drop it's time to charge it. That's all the information you need. Does anybody really think that consumers will be happy with a voltage display? I don't even know what voltage my phone operates at, let alone what the low-end of operating voltage will be.
Your Computer and Cell Phone Are Lying To You
Oh, thank God! I was worried I was the only one who could hear them!
Those who believe the Internet is private,
find their privates are on the Internet.
anthropomorphism is a common human tendency... get over it.
now if I can just get my laptop to stop humping my leg....
Use your head, can't you, use your head,
You're on earth, there's no cure for that - S. Beckett
There isn't standard regarding what reported dBm value should be associated with 1-5 bars. It is purely up to the discretion of the programmer. I have heard RSSI referred to as Relative Signal Strength Indication as well, because the value is at the mercy of internal A/D tolerances. I have seen several copies of the same radios in a lab, (Faraday Cage) report drastically different RSSI values (AKA Bars). Nearby RF sources can influence the signal levels as well.
So that part of the article is true. I dare say anyone who actually knows anything about RF won't claim, bars guarantee connectivity. To say that it is lying to you because you don't understand how it works, makes the submitter look silly. Definition of "Lie" from Wikipedia: "A lie (also called prevarication) is a type of deception in the form of an untruthful statement with the intention to deceive"
We aren't trying to deceive you, we give you the indication because it is better than nothing, and most of the time it is good enough.
It's simply Anthropomorphism. I talk to my car when it runs bad. I don't expect it to hear me or comprehend, but I do anyway. I talk to the computer, too.
When I talk to machines, for some reason it's always cursing, as in "GOD DAMNED PIECE OF SHIT..."
mcgrew's razor: Never attribute to stupidity that which can be explained by greedy self-interest
You mean that something with as many variables as strength and quality of a wireless connection can't be reduced to a value of "bars" between one and five without loss of information? Say it isn't so.
Slow news day, apparently.
It is quite common for laptop batteries to overestimate the remaining time, it even gets worse the older the batteries are: As they expose a sudden and sharp voltage drop at the lower end of their capacity, it really is hard to determine, how much time really is left.
So even though the manufacturers tend to program too optimistic parameters into the drivers, they are bound to be inaccurate as time is passing and the batteries get old.
You can use tools like IBAM from http://ibam.sourceforge.net/ to profile your batteries more accurately and gain more trustworthy readings for your time left running on batteries.
They REALLY should bring that back, Hell, all manufacturers should put that kind of thing in their cars. I can't think of a better way to make going for high mileage widely "cool". Like trying for the high score on an arcade game.
"Hey, guess what? I got 48MPG on my way to work!"
"Oh yeah? Well I got 52MPG! Beat that!"
Just simply harnessing people's competitive drive (not to mention the desire to save money) could do more than all the hybrids in the world, though that would likely lead to people buying hybrids in a quest for ever higher mileage.
upon the advice of my lawyer, i have no sig at this time
1) Cell phones.
Cell phones use a so called RSSI value for the number of the bars. RSSI is a Relative Signal Strength Indication, which is a best guess of the device how well the data transmission will go. Most use SNR directly, some use a product of signal strength and bit error rate (BER).
The reason why it doesn't always match reality is that it's really a best guess by the phone, and reality is much more complicated than just that.
2) Laptop batteries.
Laptop batteries are using charge counters. Those are resistors with very small resistance ( 0.1 Ohm) tied to a precise voltmeter in a controller chip. By integration the controller knows rather well how much charge (how many electrons) have passed through it. With Li-Ion and Li-Pol batteries in use today, however, the situation got harder because the voltage of the battery varies a lot during discharge. Nowadays, modern batteries count energy, that is the product of charge and voltage as it moves in an out, giving a very precise output of remaining energy.
The reason some batteries die very quickly once they stop showing full is because as Li-Ion batteries age, their internal resistance increases. More energy is lost within the battery during the discharge process and the amount of energy lost (and voltage decrease) is directly proportional to the current taken from the battery. At the same time, modern devices have switching regulators which take more current when voltage decreases to provide the same flow of energy to the device. Combined, this means that once the battery voltage of an aged battery starts dropping, it drops very fast.
For cell phones, this is even harder, since they don't have charge counters - the batteries have to be cheap. There the remaining energy is guessed purely based on voltage. And old Li-Ion batteries will have almost full voltage when under light load, and fail when the load is applied, causing a phone to switch off.