823247 - Wifi should not be disabled in sleep mode

Status: RESOLVED FIXED

(Firefox OS Graveyard :: Wifi, defect)

Description

[Henrik Skupin [:whimboo][⌚️UTC+2] (away 10/03 - 10/13)]

Over on bug 802418 I got the information that wifi gets disabled when the device is put into sleep mode. It's to ensure that we save the battery. But I question that this is helpful, especially when the mobile data connection is getting enabled instead. I can't say what the power consumption is for the wifi or the 3G/2G controller but from my experience with other phones and from technical reviews I got that wifi consumes lesser power than a 3G controller. If that's correct, we wouldn't safe the battery but drain it even faster.

Additionally we have to keep in mind that disabling wifi will cause the mobile data connection to be enabled. This will cause extra traffic for the user by data transfers of background applications. As result mobile users with a tiny mobile plan will earlier reach their limit, and will be dropped to 2G or will have to pay extra.

So I think that we should keep the wifi connection enabled in sleep mode as long as it is available. Once you are out of reach we can disable it and make use of the mobile data connection (if enabled by the user). Whenever you come back into the reach, wifi should be automatically reconnected even while the device is in sleep mode. 

Not sure if that can become a blocker but it should at least be discussed.

Comment 1

[Henrik Skupin [:whimboo][⌚️UTC+2] (away 10/03 - 10/13)]

Additionally I'm not sure about features of possible apps on such a phone, but if we allow background updates for apps while the phone is in sleep mode, e.g. a RSS reader downloading the full article content for offline mode reading, we would use the mobile data connection. Even harder it is if such a download is tied to an active wifi connection. That means whenever you want to update your locally cached articles, you have to turn on your phone to not exceed your mobile traffic over time.

Comment 2

[Lawrence Mandel [:lmandel] (use needinfo)]

basecamp- as this is intentional behaviour for v1. We can look to change the behaviour in a future version.

Blake - cced you in case you have any comments on this one.

Comment 3

[Henrik Skupin [:whimboo][⌚️UTC+2] (away 10/03 - 10/13)]

When do we actually check for updates? I haven't gotten a notification about an update since 2 days now given that the phone was mostly in sleep mode. Do we only check automatically when wireless is active? That would be another example why it would be important to not disable WIFI in sleep mode.

Comment 4

[Dave Hylands [:dhylands]]

My understanding is that it checks once per day.

If it's not online at the time it checks, then it sets up a listener and tries to resume when it comes back online again.

Comment 5

[Henrik Skupin [:whimboo][⌚️UTC+2] (away 10/03 - 10/13)]

(In reply to Dave Hylands [:dhylands] from comment #4)
> If it's not online at the time it checks, then it sets up a listener and
> tries to resume when it comes back online again.

Which delay will this listener have? Is it long enough to miss activity which takes about 5 minutes? I really don't get the update notification daily but sporadically whenever I didn't let my phone fall into sleep.

Comment 6

[Rubén Martín [:Nukeador] ❌ [away till Aug 31st]]

+1 for not disabling wifi when screen goes off, it's really annoying. This should be an opt-in preference under wifi settings.

Even updates are interrupted by this behaviour and a lot of battery is drained switching to 3G and back to wifi+connecting AP...

Comment 7

[Vincent Chang[:vchang][changyihsin]]

Bug 859895 - Make wifi timeout build-configurable is landed, we can disable wifi automatically turn off by modifying wifi.screen_off_timeout to 0 in gaia/build/settings.py.

Comment 8

[Rubén Martín [:Nukeador] ❌ [away till Aug 31st]]

Hi,

I was trying a chat app on a Keon device, and with this behaviour is not possible at all to run an app that should be online all the time.

Shouldn't this be blocker to allow these kind of demanded apps be functional on launch?

Comment 9

[Henrik Skupin [:whimboo][⌚️UTC+2] (away 10/03 - 10/13)]

AFAIK we don't block on basecamp anymore. Moving to leo.

Comment 10

[Wayne Chang [:wchang]]

ni? product as current behaviour was previously decided.
Need product to comment if this needs to be reviewed so that Wifi is kept enabled when device enters sleep.

Comment 11

[Nikhil Marathe [:nsm] (No longer reading bugmail, please needinfo?)]

For Push notifications, we'd like to keep a connection open for as long as possible. When wifi is disabled on sleep, push has to switch to 3g to receive notifications. This is not a very large data hit when there are no notifications. I've no idea about relative power consumption. On the other hand, if the user is in a place with no 3g/data connectivity, but is connected to wifi, disconnecting the wifi means that push isn't really push anymore. The network stack does not detect when a socket open is attempted. At the least wifi should reconnect if some process tries a network operation.

The other annoying thing when 3g and wifi both are available, with respect to push, is:
1) User connected via WiFi, push socket is over WiFi.
2) Phone goes to sleep, socket switches over to 3G.
3) User gets a notification, switches on display.
4) Sigh... push socket switches over to WiFi again.
This cycle will continue.

Comment 12

[Rubén Martín [:Nukeador] ❌ [away till Aug 31st]]

Also switching to 3G is a lot more power consuming than Wifi. This means if a leave the phone at night at home, it's going to use my 3G data and more battery.

Comment 13

[Alex Keybl [:akeybl]]

Product - please renominate if you feel this should be a blocker for a future release.

Comment 14

[Peter Dolanjski [:pdol]]

We will consider this for Koi.

Comment 15

[Shian-Yow Wu [:swu]]

This can be solved by build time configuration in comment 7.

Do we need it to be run time configurable?

Comment 17

[Preeti Raghunath(:Preeti)]

UX,

Please comment

Comment 18

[Stephany Wilkes]

What is the UX need here? Please clarify and we'll triage. Thanks!

Comment 19

[jachen]

Hi there-
I'm going to assign needs info to Mike in Taipei. He can help to work with Carrie and Juwei to get feedback.

Thanks!
Jaime

Comment 20

[Neo Hsieh [:neo]]

Hi All, Can anyone prove that ? 
(The WiFi consumes lesser power than a 3G controller)

Comment 21

[GoodMike]

Excuse me, Are we talking about sleep mode = suspend ? I don't think we have sleep mode or power saving mode in our device. This can cause different solutions and discussions. Thank you.

Comment 22

[Henrik Skupin [:whimboo][⌚️UTC+2] (away 10/03 - 10/13)]

Isn't it called sleep mode when you turn off the screen via the power button? That's what I got from various other bugs.

Comment 23

[Preeti Raghunath(:Preeti)]

Peter,

What is the expectation here with Wifi. This seems to be going in circles and we need guidance here.

Comment 24

[Ken Chang[:kenkai]]

(In reply to Neo Hsieh from comment #20)
> Hi All, Can anyone prove that ? 
> (The WiFi consumes lesser power than a 3G controller)
Theoretically, the wifi component consumes lesser power than a 3G radio component in "connection" mode.
But it is better to directly measure the power consumption of device.

Comment 25

[Preeti Raghunath(:Preeti)]

IRC with Pdol this is a feature where requirements are getting spec'ed. So agree with 1.3?

Comment 26

[Peter Dolanjski [:pdol]]

Sandip is checking with the RIL team on the possibility of addressing this in 1.3.

Comment 27

[Ken Chang[:kenkai]]

I wonder if we can handle this issue. In current OEM design, wifi can not enter the deep sleep mode. That means the WIFI has more power consumption than 3G data call during the sleep mode.
If we really want to have keep wifi on and have the better power consumption during sleep mode,
we need OEM's support to have a wifi chip/component enabling deep sleep mode.

Comment 28

[Sandip Kamat]

(In reply to Ken Chang from comment #27)
> I wonder if we can handle this issue. In current OEM design, wifi can not
> enter the deep sleep mode. That means the WIFI has more power consumption
> than 3G data call during the sleep mode.
> If we really want to have keep wifi on and have the better power consumption
> during sleep mode,
> we need OEM's support to have a wifi chip/component enabling deep sleep mode.

Can we just provide the option to enable/disable wifi during sleep mode for v1.3? We can have OEM make the decision of changing the defaults (and further optimizing the power consumption later).

Comment 29

[Ken Chang[:kenkai]]

Have commented in
https://bugzilla.mozilla.org/show_bug.cgi?id=930972#c5

Comment 30

[Jason Smith [:jsmith]]

This is being addressed as part of the feature in bug 930972, so I'm duping over there.

Comment 31

[Ken Chang[:kenkai]]

It is not a duplicate of bug 930972, but a blocker for 930972.

Comment 32

[Jason Smith [:jsmith]]

This doesn't fall under the QC blocking feature list & DSDS feature list. Renoming.

Comment 33

[howie [:howie]]

Hi Ken, could you help to assign this?

Comment 34

[Ken Chang[:kenkai]]

After fixing bug 979130 and 960623, we don't need to fix this. Everyone who want WIFI to keep waking up just can use that wifilock mechanism.

Comment 35

[Henrik Skupin [:whimboo][⌚️UTC+2] (away 10/03 - 10/13)]

(In reply to Ken Chang[:ken] from comment #34)
> After fixing bug 979130 and 960623, we don't need to fix this. Everyone who
> want WIFI to keep waking up just can use that wifilock mechanism.

My original request with that bug was to not get wifi disabled during sleep mode. With the bugs you pointed out here, this got fixed. So not sure why you set this bug as wontfix.