Battery thief framed by app: all details inside

Following up from a battery draining problem, I've found a tool (BetterBatteryStats) that profiles battery usage in grater detail than the Android stock battery tool:

http://www.howtogeek.com/139592/how-to-find-the-root-cause-of-your-android-battery-problems/

Here is the profile of a full day plagued by anomalous battery consumption:

BetterBatteryStats has found RILJ to be the prime suspect:

It seems to be a known behavior as said here:
http://forum.xda-developers.com/showthread.php?p=31352333

http://forum.xda-developers.com/showthread.php?t=1878692
http://forum.xda-developers.com/showthread.php?p=31352333



I've followed the instruction and disabled Google geolocalization:

After some time running, RILJ is still top in battery usage, but it's weight is comparable to other processes, and battery consumption is a reasonable 1%/h

All went fine for some days, when after sending an SMS and receiving a call, the issue surfaced again.

http://code.google.com/p/android/issues/detail?id=17383

I tried disabling "increase volume when in pocket".

My phone is running fine since then.

Who's stealing my battery?

My Galaxy S2 Plus always had a problem with battery consumption.

This is a normal daytime battery usage:

Then, sometimes out of the blue, battery usage increases drastically:

I still haven't found out the root cause.

I even upgraded to Android 4.2.2, with no luck: the issue still persists.

Here is the usage breakdown: what's causing Android to use more juice than the phone and the screen, usually the most power hungry subsystems?

I used CPU Spy, but it doesn't seem to be an issue of deep sleep.

Motorola Defy vs Samsung S2 Plus

I've been using a Motorola Defy for the past 2 and a half years.
Now I switched to a Samsung S2 plus, expecting to find some improvements.
The S2 has a bigger screen than the Defy, but a lower resolution. Anyway I like the bigger screen more than the lack of resolution, that is not dramatic.
The AMOLED screen is way easier to read in sunlight, but contrast and gamma seems out of range: reds are loaded and dark photos are darker than on the Defy.
Current Samsung firmware doesn't have color calibration either.
With some gamma correction, this screen could kick asses.
With a bigger screen I would have expected a better on screen keyboard experience, but typing on the screen is a pain as always.
The battery is another notable difference from the Defy. It just drain faster on the S2. 
Way faster.
I never had to charge the Defy in the middle of the day. I wish I could say the same of the S2 Plus.
I don't know if it's a hardware issue or a software setting, but I have the clear feeling that the lower part of the screen is less sensitive.
The Defy was resistant to water and dust. The S2 clearly is not. This is something that I miss.
The S2 has double the RAM and double the cores of the old Defy.
This makes the S2 more responsive, but I expected a little more: there are still some apps that force the launcher to be reloaded, like dolphin browser.
The GPS gets a fix in a handful of seconds, probably thanks to the glonass system, but the compass is always wrong.
Even after calibration. This makes using Google sky really frustrating.
In the end, while the S2 plus is a good phone, I still think that the Defy, at its time was a notable terminal. An updated Defy would be a clear winner for Motorola.

Android Activity Recognition Battery Usage

I'm using an amazing android app called AutomateIt.
It's kind of a scheduler on steroids: it can execute actions based on triggers like time, position, connection status.
The latest release added support for "activity recognition": the phone should be able to tell if you're walking, driving, or standing still.
I defined a rule that turns on Bluetooth while I'm driving, so I can use my car hands free system.
It actually works, but at what price?
Here are the effect on a Samsung Galaxy S2 Plus.
This a night with an activity recognition rule defined: check the last hours in the graph: battery depletes at a rate of 1% every 4 hours.

Here is a night without any activity recognition: a 1% of battery juice lasts for 5 hours.

It seems that using activity recognition shorten you battery running time by 20%.

Comparison of GPS performance across different Android devices

I had the chance to test GPS on a Galaxy Mini 2 and a Galaxy S2 Plus.
The test was done in the same place, one phone after the other.
Both phones didn't have any data connection, cellular of Wi-Fi, so positioning was done using only plain GPS.
Here is the Galaxy Mini 2.
Locking took 60 seconds with 5 satellites.

The Galaxy S2 Plus got its lock in 3 seconds with 16 satellites.

Note that on both phones, compass was way off.