What is the best way to model ham DMR repeaters?

[unusedusername]

Hello All,

This is a data modeling question regarding Ham DMR repeaters.

I already have a script that I use for my scanners that takes a few different repeater DBs and generates a set of three (West, Central, Eastern) hpd files.

I can't share these lists, unfortunately, as the data providers all have policies that the lists are for personal use only. The data isn't very "clean" either as the lists have many duplicate entries, and the range circles are sometimes pretty far off from the reality of how far the repeaters reach. It's good enough for me to play with though, and I enjoy the challenge of programming the radios.

I'm thinking of doing the same script based generation for DMR.

I can get the list of all 2216 DMR repeaters worldwide that are linked through the standardized talk groups from dmr-marc.net. I can also get the list of ~38,000 valid DMR IDs for the standard talk groups from the same location. I can also get the list of the few hundred coordinated talk groups.

By filesize, I should be able to fit all the DMR repeaters in the US, and a translation from DMR ID->callsign, and a list of all the standard talk groups into a single hpd file, generated by a script.

The question is how do I want to model it.

Here is what I can see as possibilities

1. Conventional System with DMR frequencies
   This is relatively easy, in fact I've already done it. The downside to this is that the scanner does not display who is talking, nor the talkgroup or channel being used even though the information is available on the digital data stream.
   
   
2. One giant system per state
   I could add one giant system per state, where every repeater site in the state, every valid talk group (~500?) and every valid user (38,000) is entered into the system. I'm not sure what system type to use here.
   
   The main downside is that I would be repeating the user list of 38000 IDs once per state.
   
   The advantage is that I should be able to see who is talking, and on what channel.
   
   
3. One giant system per region of the country
   This is the same idea as before, but expand the per-state idea into three systems covering the US so I don't need to repeat the ID list very often.
   
   The downside is that systems in ARC536 have a limit of 999 sites. After that the software crashes. I won't have to worry about it now as there are only ~2200 repeaters world wide, but I would have to worry about it when a single region of the country builds more then 999 sites with repeaters. I can do the same trick I used for conventional systems and make a "State-1" and a "State-2" system when I get beyond 990 sites.
   
   The upside of this is that I will only need to use the ~380kb of user data once per region instead of once per state. I would also be able to see who is talking to who, and on what channel.
   
4. Something else?

Some questions:

Is there a better way?
What type of DMR system would I need? (Single frequency?)
One assumption here is that all coordinated repeaters would be using the same talk group IDs to represent the same group. I'm not entirely sure this is the case.

I'd love to brainstorm ideas on how I should put this together.