Exactly this setup worked for me; albeit I have chosen another installation method:
Good news: The legacy version of µg Unified Nlp is now available from F-Droid. This makes installation easier.
Easier how? Is it not necessary to make it a system app anymore? On F-Droid’s App description it sounds like it’s used as a user app, but it also says it’s for AOSP devices.
Yes, for me it works without making it a system app.
Thanks, works for me to 
I updated the original post.
OK, thanks to the new F-Droid package I was able to install Unified Nlp, and by using the LocalGSMNetworkBackend (downloading the database filtered by my GSM provider ID) and Nominatim Backend I’m able to get instant location fix on OSMand even inside my office without any satellite visible!
Does anybody use µg Unified Nlp alongside Google Services? If so we could add an Installation Guide for those users.
They’ll obviously have to install the µg Unified Nlp for devices with GAPPS and they’ll have to install the Xposed Unified NLP Module.
I’m trying that, Paul.
Not starting really well, actually:
I’m not sure what to do next, except turning on location access.
That’s part of my problem to. I keep getting a (de)install error.
Version 7.0.99 (1809214-034 is installed, but greyed out in the installed F-droidlist.
I can’t remove it.
Version 1.1.10 is also installed. but won’t work because it’s signed with another key.
I think thats normal.
Please excuse my bad tech-language but I believe Unified NLP for GAPPS-free devices uses the same package name “com.google.android.location” to let Android believe that Google’s Location Services are installed.
Also “ichnea” is a word that Mozilla uses for some of their package names.
So neither of theese should be reasons for this not to work.
Also now I’m wondering, how would you even know if Unified Nlp is working, since you should get a Location from Google anyway? 
Hm, when I started using my FP, I tried for a while to go without GAPPS, and fumbled with NOGAPPS (check this thread on XDAdev). Maybe that’s a legacy of this fumbeling? Did you try that too, @Lidwien?
An @paulakreuzer’s comment is of course right: how would I know that it works? 
BTW, I used Nokia’s server to get a lock in WiFi locations when I wasn’t using Google, which worked quite well. But AFAIR, cell tower based location did not work at all.
I started with Google maps and then OsmAnd. Google Map is a disaster because you can only download parts of a country instead of a whole country. And you can’t use it offline. I had problems with the Gps, which I solved by installing FP1-EPO-autoupdate. I never tried CyanogenMod, so I don’t think I ever used NOGAPPS. If GAPPS comes with the Fairphone OS, then I might have used GAPPS.
OT, cause it’s Friday:
I just looked that up. Come on, bugger off, I thought! 
Ichnea, a genus inside the Cleridae? Mozilla must have confidence in their packages to name it like that.
Clerids have a minor significance in forensic entomology. Some species are occasionally found on carrion in the later dry stages of decay. Also, some species are pests (stored product entomology) and are found infesting various food products. Research efforts related
to Cleridae have focused primarily on using certain species as biological controls. This is a very effective technique for controlling bark beetles due to the voracious appetite of many Clerid species.
But then I realised the spelling error: it’s ichnaea, not ichnea. A much better choice of name, from greek mythology. It might still refer to some bugs. Ok, it could have been a nicer bug in the ichnea spelling as well, and don’t worry, there are also loads of entomologist who can’t spell their greek and latin properly. Ah, well, it’s a bugs life, anyway. 
Actually I was wrong. I confused “Ichnea” with “Fennec”. That’s what Mozilla Aurora’s package name is called. “org.mozilla.fennec_aurora”
My bad
Yep, it’s the codename for the location service only. (I linked the github repos above.) One of the main contributors wrote an interesting blogpost about Mozilla’s ichnaea in 2013, btw. The documentation is also fascinating, e.g. the point “calculation”.
Some functionality of µg Unified Nlp is also provided by GMS core, which is also part of the µg project. I have not tested it, however.
I have written a post about GMS core here:
I finally managed to compile and flash the FP Open Source OS to the FP2. I used f-droid to install the µg UnifiedNlp (no GAPPS). No I’m wondering which backend(s) I shall install. f-droid offers different nlp backends. I would like to choose the one with the best (up to date) coverage (in Germany), preferable working offline. If I understand it correctly, I can install more than one as well…?
Anybody good/bad experiences with a specific backend? Which one would you recommend?
So I assume you have also added the needed patches as well?
You want offline, so I would recommend using GSM (cell-tower) information (like GSMLocationNlpBackend and others). Not sure how good they are for your area, let us know 
If you live in a big city, adding WiFi APs might work as well.
Yes.
Not sure if I made myself clear enough. I intend a use-case where I only need sporadic internet access (via WiFi) to initialize/update a local database and then can use the service without internet connection. It would be good to have data for GPS/WiFi/GSM to have all fallbacks available, depending on which information is present at a current location.
Well, about 35,000 inhabitants, so rather not too big. 
But I’m using Mozilla Stumbler to collect data. 
Hard to tell, depends on your use case I guess. I just use my GPS, if the data helps me to get a quicker fix … good. So if you do some mapping and need good fast GPS data for that, create a WiFi AP map for the area first so you get a quicker GPS fix later. Not sure if really needed, though. Also there is lot of better hardware out there to get very good GPS fast … you just need to download some data before you enter the area you want to survey.
Most of the time I want the opposite: My phone should not tell everyone where I’m am. So it’s best if it thinks I’m elsewhere
