[vueko]

While definitely better than the non-encryption of CTCSS etc, P25 encryption has some relatively concerning implementation problems. It's possible none of these actually matter for your usecase as many are related to UX around configuring and using encryption in an institutional setting, but the unauthenticated traffic injection, induced transmission and jamming issues are, well, not great no matter how you look at it.

https://www.mattblaze.org/blog/p25

https://www.mattblaze.org/papers/p25sec.pdf

I will grant that the open-source kfdtool keyloader boxes are neat.

I have been meaning to see if I can repro the induced transmission via retransmission requests thing when the data packet stuff is fully disabled via CPS, but a friend permanently borrowed my hackrf so that project is on hold for now. I'm not optimistic, though, due to the comments in the paper about where in the stack the retransmission request is processed.

[Aloha]

A lot of these issues if not all are specific P1 P25, which if you're on conventional you're using. I find it interesting they treat P1 as if it's different from normal FM, it is not, its C4FM - so all the normal capture effect stuff works fine.

In the end, the goal of P25 was to make it almost, but not completely possible to listen in - I'd also note the way they did their testing is not how these radios are typically deployed, which is not direct subscriber to subscriber on simplex, its either via a conventional repeated system or via a trunked system.

I do agree that frequent key changes are bad - but for entirely different reason, they're an operation nightmare. Most agencies who deploy it either use very long lived keys deployed by KVL, or deploy a UKEK and then change keys via OTAR on some cycle, which is not infrequently, never.

My belief is the theoretical risks on P25 are not well proven in reality, you're more likely to have some interference issue over say, intentional jamming.

Either way a radio for ~75 bucks plus supporting hardware that will do AES256 isnt bad - and DES-OFB/DES-XL is good enough to keep the casual listener out, which is more or less the point of the crypto here. Also I know they make a big deal about unit ID's, but they carry no meaningful info, its just a hex ID.