SMART DIGI ALGORITHMS 31 OCT 2003 ------------------------------------------------------------------------- Updated 13 Dec 2004 to the new n-N paradigm WB4APR Updated 11 April 2006 There has been lots written about smart digi algorithms, but I have serious doubts about their efficacy for two fundamental reasons: 1) Smart digis, by definition, make the network undeterministic. One has no idea what the network will do with your packet. Will it go where you want it to go? Does the local sysop's opinion of where your packet should go match with your need? Does the local digi even understand your smart request? 2) With over 1,000 digis in the USA, and almost as many different digi sysops and opinions and beliefs and personalities, my experience has shown me that the hope of getting any consistent implementation of even the simplest of smart digi concepts would be very sparse and spotty and would take years and years and years to even get to half... What good is a smart network that is half dumb and you the user can never tell what the half smart and half dumb network will do with your packets. See #1 3) Almost all of the "problems" that were in such dire need of fixing have mostly been fixed with the 2004 initiative called the New-N Paradigm. Those simple fixes using existing digis and no requirement to change out any hardware made a DRASTIC improvement in most areas. I have seen reports of reductions by a factor of 3 to 5 or more in the amount of QRM on the network. As a result, the network is now (2006) much more reliable for local use. In conclusion, I am quite opposed to digis "making assumptions" about what a user wants his packet to do. I believe the USER should always be able to choose his own best path and have the network respond to it. This is the only way to have a deterministic network which operates consistently everywhere for the user. Of course, the network is allowed to protect itself from abuse also. SMART DIGIS. But since this concept seems to come up often, I will try to document the various proposals. So here are two lists. The first is the list of smart-digi directives and the second is just a list of the commonnly used callsign-substituting aliases that may be in use in some areas... Send me more if you think of them... SMART DIGI DIRECTIVES: Directives are pseudo-path names that are inserted on the end of an APRS user's outgoing digipeater path to serve as a signal to the network of any special handling requirements. Here are several of the ones that have been suggested: NOGATE - IGates should ignore these packets RFONLY - IGates should ignore these packets (why do we have 2?) IGATE - Smartdigi may substitute shortest path to nearest IGate ISTATE - Smartdigi may substitute a ##LNKn-N along an interstate STATE - Smartdigi may substitude the local state SSn-N LAN - Smartdigi may substitute local LANn-N name LINK - Smartdigi may substitute an optimum LINKn-N path EOC - Smartdigi may substitute shortest path to nearest EOC SKYWRN - Smartdigi may substitute preferred skywarn path WX - Smartdigi may substitute preferred local WX path NOAUTO - Smartdigi should do NOTHING to this path CALLSIGN SUBSTITUTING ALIASES and TRAPS: In early APRS, callsign substitution was used to provide some traceability to the paths taken by the GENERIC path callsigns of RELAY and WIDE. But these generated many duplicate packets and were phased out completely in the 2004 New-N Paradigm. Now, callsign substitution is used as a TRAP for abusive paths. It works as a trap, because the matching path is digipeated but replaced with the digi's substituted callsign. This effectively stops all further propogation of that path if it was a WIDEn-N path. THe four traps typically used for a New-N Paradigm Digi are WIDE7-7, WIDE6-6, WIDE5-5 and WIDE4-4. STATE AND SECTION PATHS: Originally there were some special aliases that were used to support local nets such as SS, SAR, SKY, ARES, RACES, etc. But these also had the same DUPE and flood problem of the original RELAY and WIDE. Hence these were all also obsoleted with the New-N Paradigm. They were replaced with the nomenclature SSn-N. Where SS is the abbreviation for a small state or for the ARRL Section in a large state. SSn-N packets propogate just like WIDEn-N packets, except that they stop at the borders. THis way, their damage to the network outside of their intended area is completely eliminated. For example, a WIDE6-6 packet launched in Maryland would potentially QRM over 144 surrounding digis in over 13 states. But a MD6-6 packet during a routine State Net, would hit only a total of 9 digis (the 9 big digis that cover Maryland) and would go no farther. And it would work the same for someone in the middle of the state or at the extreme pan-handle. Since SSn-N packets do not get traced except for the last hop, an improvement is to actually use WIDE1-1,SSn-N. This takes advantage of any fill-in digis and also assures the First and Last digi paths are traced. This way the packet will arrive as DIGI1,WIDE1,DIGIn,SSn-N with both the FIRST and LAST digipeaters identified. de WB4APR, Bob