3G Watchdog app for Android monitors your 4G/3G/Edge/GPRS data usage

GB, MB, and KB data measure units

Written by Richard. Posted in FAQs

Officially the GB, MB, and kB unit symbols are based on decimal arithmetics (1 GB is 1000 MB, 1 MB is 1000 kB and so on) whereas the GiB, MiB, and KiB are based on binary calculus (1 GiB is 1024 MiB, 1 MiB is 1024 KiB, etc…).

In practive however this standard is largely ignored. All ISPs use GB / MB symbols but for some 1 GB is 1024 MB and for others it is 1000 MB. 

This is very confusing but unfortunately there is no easy solution. 3G Watchdog uses binary  arithmetics (1 GB = 1024 MB, etc…). If you find out that your ISP is using decimal arithmetic (1 GB = 1000 MB, etc…) you must adjust the value entered in 3GW for your quota as follows:

value entered = (ISP value in GB or MB) x 0.9765

Usage prediction [Pro]

Written by Richard. Posted in FAQs

3G Watchdog Pro usage prediction capability is one of its most useful features. It is important for the user not only to know her/his current consumption, but also to anticipate its usage by the end of the current plan period or day so she/he can take corrective action if necessary.

However usage prediction is not an exact science, and there is no way to predict usage with 100% accuracy. Different strategies may be implemented, and as of version 1.22.0, 3G Watchdog offers the choice between 3 prediction methods:

  • Standard (aka “Amnesic”) method:  this prediction is based on usage since the beginning of the current period (day or plan), and is a simple linear extrapolation of this usage. Usage history before the beginning of the current period is not taken into account (hence the nickname “amnesic”). A drawback of this method is that the prediction is unavailable at the beginning of each period until enough usage data can be collected to get a significant prediction. Even after enough data has eventually been collected to supply a prediction, this prediction strategy still tends to amplify your usage in the initial part of the period: for example if you had exceptionally high consumption in the first hour, the prediction will be very high too, which can be disturbing. Nevertherless the prediction will become more smoothed and accurate as time passes.
  • Moving average [since 1.22.0]: this prediction is based on the average usage from the last N days (N configurable, default: 30). This strategy tends to smooth the prediction because past usage (before the current period) is taken into account. It works well on a large period such as a monthly plan, but may produce strange results with current day prediction, as the importance given to past usage may outweight the actual usage for the day.
  • Mixed [since 1.22.0]: this prediction is based on a dynamically weighted combination of the two above methods. More weight is given to the moving average prediction (based on past usage) at the beginning of the period (reflecting the absence of actual data), but as time passes and more actual usage data becomes available the balance shifts towards the amnesic prediction (reflecting the fact that actual data is more important than speculative one).

No method is perfect. You might want to experiment in order to find which one best matches your usage pattern. You can freely switch from one method to the other. Go to 3GW Settings > Prediction to select the prediction method.


Incorrect usage count on the Galaxy Tab 10.1, 10.1N, 10.1v (and possibly 8.9)

Written by Richard. Posted in FAQs

The Galaxy Tab 10.1, 10.1N, 10.1v, and possibly the 8.9 don’t work well with data usage counting applications. 3GW leverages functions provided by the Android programming interface, and apparently they are incorrectly implemented on these devices (see Android issue 19938). There is nothing we can do about it (we have no access to the faulty code). Let’s hope that Samsung will fix this in future upgrades of Android. Sorry for the inconvenience.

The Pro version since 1.10.0 has a workaround: you can check the “Calculate global usage from usage by app” option in 3GW Settings > advanced to tell 3GW to use the sum of all application usages as the source for global usage. It’s not a perfect solution though, as usage by app is generally less accurate than global usage.

Update Sep 15, 2011: if you are ready to take the risk of flashing a new kernel (and assume the consequences), custom kernels available on this page, from version (Aug 26, 2011) or more recent, allegedly fix the problem for the Galaxy Tab 10.1 (not verified though). Instructions to flash the kernel are included in the page. You also need to uncheck the 3GW Settings > Advanced > “Use pre-Froyo traffic counting interface” option for the counting to be correct.

The usage % calculation is inaccurate?

Written by Richard. Posted in FAQs

The usage % calculation is bogus” – (variant of “The % is stalled at 0 [since the last upgrade or other conditions”)

Unit symbols used in computing are misleading, because a K is not 1000 but 2 power 10, which is 1024, so 1KB is 1024 bytes, 1MB is 1024 KB, and 1GB is 1024 MB. For example for someone with a 1GB quota and 112.04MB usage, the ratio is 112.04 MB/ 1024MB (10.94%), not 112.04MB/ 1000MB (11.2%).

The % is stalled at 0 [since the last upgrade or other conditions]

Written by Richard. Posted in FAQs

Check that you didn’t accidentally select GB (GigaBytes) instead of MB (MegaBytes) for the Quota in the Settings.. page (for example a quota of 500 MB would make sense but not 500 GB). Since 1 GB = 1024 MB, that would explain that the percentage looks like it is stalled at 0.

Is usage by application accurate? [Pro]

Written by Richard. Posted in FAQs

Is usage by application accurate [version pro, Android 2.2+]?

Usage information is provided by the Android system, and unfortunately is rather approximative. As a rule of thumb global usage (as displayed in the main screen) tends to be more accurate, so you should trust it more.

Starting from 3GWP 1.9.0, streaming media traffic (e.g. YouTube) may be reported under the name “Streaming media” rather than under the apps which actually use it. However the usage of some applications (e.g. TuneInRadio) is correctly reported under their own name.

Measured usage not assigned to any specific installed application is listed under the name “OS services + others”.

The lack of separate counters for Mobile and Wifi usages obliges 3GW to “track” connectivity changes to distinguish the current type of usage. This creates a risk of overlooking usageoccurring while the 3GW service is not running. The service is automatically started on boot and normally stays running in the background,  unless killed (manually or automatically) by a task killer, or by the system in case of a severe shortage of resources . If this happens, the service will automatically be restarted within 15 mn, but traffic by app consumed during that time will not be counted. Note that the same problem will not affect global usage because Android provides a dedicated counter for global mobile usage, so even if the 3GW service is temporarily down, the internal Android counters are still tracking usage and the 3GW service is able to catch up without loss of data when restarted.

The sum of all usages is displayed at the top of the screen can only be compared with the global measured usage, not with the billable usage.

Finally, it has been reported that the Dell Streak and Acer Mini E310 don’t implement correctly the Android functions used by 3GW, and usage reported stays at 0 for all apps.

Why are usages by application sometimes shown in blue?

Written by Richard. Posted in FAQs

The blue color is just a visual hint to highlight usages that have changed since the last refresh. It will disappear with the next refresh, unless the usage has changed again.

Can 3GW exclude certain types of usage (TV, sites, applications…)?

Written by Richard. Posted in FAQs

In the Pro version 1.6.0+ applications can be excluded (i.e. not counted) in the usage by application total, but since usage by app accuracy is dubious this solution is not entirely satisfactory.

Filtering by destination IP is hard since it would have to be implemented at the system level, like a firewall, and would require a rooted phone with access to the IP tables (which are not present on all phones)!

Can I have a longer (one day) update period?

Written by Richard. Posted in FAQs

(usually in order to save the battery). Currently the available update periods range from 30 sec to 1 hour. With an hourly period, 3GW runs 24 updates per day. Each update takes around 150 ms (clock time, not CPU), thus ~3.6 seconds per day of processing clock time. This will certainly not drain the battery or bring the CPU to its knees. So there would be no benefit to increase to period beyond 1 hour. Besides, the 3GW widget and notification bar refresh depends on this update, and 1 hour is already a lot.

A final important reason: when the phone is rebooted (turned off/on), it resets its internal counters, and there is a risk that the usage since the last update be lost/not counted (3GW tries to detect this situation and work around it but it’s not guaranteed to work at all times). The higher the update period frequency, the lower the risk, so a high frequency is recommended if accuracy is important.

Follow Us