µg - A plugin-based replacement for gapps (updated)

This is a follow-up of Nogapps - A replacement for gapps.

The NOGAPPS project has been renamed to µg, and the part that interessts me the most, off-line, libre, coarse localization is called µg Unified Nlp.

The forum is sparse on usage information, so this aims to be a how-to with screenshots.

Prerequisites

  • A rooted phone
  • No Google services installed
  • Android 4.4 - 5.0. If you’ve got and older version, see the projects github.

Installation

  • Install µg UnifiedNlp. Installing from f-droid is the easiest way.
  • Now you need providers. Search for nlp in f-droid, and install i.e. GSMLocationNlpBackend. This will give access to two databases of GSM cell towers and their locations.
NLP

Configuration

  • Find the “µg UnifiedNlp” app and open it, go to “Configure location backends”.
unifiedlocation
  • Hit the gear to configure GSM Location Backend.
gsmlocation
  • Mozilla Location Services can be activated right away.
  • To activate OpenCellId you need an API key. Head over to OpenCellId.org and register.
  • Make sure you’re on a decent network, preferably WiFi, and hit “Generate Database”. It will download a few hundred MB, depending on your location. Note that one seems limited to one database dump from OpenCellId per day. Subsequent attempts will fail.
  • You also have to make sure your android location settings are set to “Powersave” (WiFi and GSM) or “High precision” (GPS + WiFi and GSM).
positionsettings

Backend alternatives

Currently a search for Nlp in f-droid gives,

There seems to be more Network Location Providers on the way, like OpenWLANMap.

Testing

Now does this work?

Using LocalWifiNlpBackend I can find my location without GPS, fast. But this is a very selfish method (not sharing data), and it only works where your phone has been and had GPS signal before.

I’ve had less success with GSMLocationNlpBackend, even though I know my neighborhood is mapped. Will update this if I experience differently.

noposition

OpenWLANMap is probably a better bet for proper localization, so hold on for those to arrive!

Edit: After the update of µg UnifiedNlp (Version 1.2.0 - Added on 2015-01-19), localization works great for me! (And by that I mean the test that failed in the last paragraph).