At 06:59 AM 3/10/2010, you wrote:
>I have been doing a lot of thinking about a new D-STAR Architecture >(and hope to provide some more concrete material at some point), but >this discussion brings up some basic ideas that have been >percolating in my ruminations. > >Icom started from a different model than community usage patterns >suggest is the proper way to do things. There is certainly a mismatch between the Icom implementation and the typical usage patterns. >Here are some of the design principles that I think are important: > >1 - Internet connected devices need an easy mechanism for discovery >- use Secured Dynamic DNS. (Set TTLs based on likelihood of IP >Address changes (static IPs have very long TTLs, dynamic TTLs very short) Yep, good start. >2 - Have a low latency, redundant and reliable service to map >radio/repeater/simplex node/gateway/service Callsign Addresses to >DNS names. Maybe something like a Cassandra (Apache) cluster. (Use >caches on local boxes for the "working set of callsigns" that a >local system needs, don't copy full databases around.) Well, systems like Echolink manage to avoid a lot of lookups. Echolink does update the servers, but at least some clients cache the IP list, and most updates are partial. >3 - User authentication - completely eliminate registration for RF >devices that use Callsigns, since Callsign Addresses are sent "in >the clear" anyone can forge them (pirate the callsign) and Agree totally. I can't see a valid reason for RF registration, at least for DV mode. At the same time, design in a strong authentication system for direct Internet access (DV Dongles, D-RATS, etc), and back that up with a validation process that is as thorough as is practical (in the real world). >deterrent than a silly "registration" system. A fairly simple >regular expression could filter "MY" callsigns to make sure they at >least look like a legitimate callsign, here is a Java Example/Test >code for callsign checking, including a mark for VK foundation >licenses (who are not permitted to use D-STAR): Yes, that filter could be used in a few ways, from refusing to forward traffic from these callsigns. However, this sort of filtering would actually be best applied at the repeater. Echolink has a similar policy. They won't validate VK Foundation calls for link or repeater stations (but will for direct PC access), because they have been informed that "F calls" are not allowed to setup RF gateways or repeaters. They are allowed to _use_ other peoples' RF (on FM, not DV of course), or access Echolink from their PC directly. One quirk of our regulations is that it _may_ be legal, or at least not considered worth bothering with, for a Foundation call to access D-STAR gateways outside Australia using a DV Dongle and no RF locally. Certainly no one has raised any eyebrows about F calls accessing 23cm, 6m or even 220 or 70 MHz repeaters in foreign countries from their PC. Technically, it's an "elevation of privileges", which is an issue with our laws, but as no RF is generated locally, then the ACMA has no jurisdiction, at least as far as the amateur regs go. The advent of Internet connected modes means authorities _really_ need to rethink how regulation works. 73 de VK3JED / VK3IRL http://vkradio.com
