[aprssig] Premptive Digipeating (b) PATH MARKING

Robert Bruninga bruninga at usna.edu
Fri Nov 30 12:35:59 EST 2012


Taking comments to date, I added the sentence about keeping the remainder
of the path open, and attempted to address Steve Dimse's very valid points
about traceability.  I'm suggesting we can use the RR bits in the SSID
field. CAN WE DO THIS?

PREMPTIVE DIGIPEATING DEFINITION (rev b):  When ON, the digi will search,
starting at the first unused field and then to the end for a match to any
of its potential matching aliases (MYCALL, MYALIAS, UIDIGI).  If a match,
it will mark that and all prior fields as HAS-BEEN-USED, it will REPLACE
its ALIAS with its MYCALL and will digipeat it once.  All remaining fields
are still active.

INDICATORS:  When the packet is preemptively digipeated, the preempted
path can either be DROPPED, or MARKED.  If dropped, then the packet
arrives with only the preemptive digipeater's MYCALL and remaing path as
normally done.  All prior path data is lost.  Or if MARKED is SET, then
the entire path is retained but the LSB of the two RR bits is set to 1
indicating the field was prempted.

In the GATEWAY scenario, one would probably use the DROP setting.  In the
normal network scenario, then the MARK setting might be the best setting.

PRIORITY BIT:  Now why not use the other RR bit in the SSID field for a
PRIORITY bit.  The existing default value of the RR bit is 1 and will
indicate a PRIORITY  packet but if "0", then this packet is of less
priority and can be dropped in future networks that recognize it..

QUESTION:  Does playing with the RR bits break any existing processing?
To see the definition of these bits, see figure 2.2.13.1.1 in:
http://www.tapr.org/pub_ax25.html

So the PREMPT TNC command now has 3 settings:  OFF, DROP, MARK to indicate
how to indicate the action.

The rest of the below remains as originally
proposed:-----------------------

PACKETS TO HOME SCENARIO:  In this scenario your goal is to get your
packets targeted to a particular digi or area.  The path WIDE2-2,HOMEX
With the HOMEX digipeater recognizing pre-emptive digipeating would make
sure that all your packets got to HOMEX in from 1 to 3 hops.  To get them
say 4 hops from City A to City B without flooding everywhere, the path
WIDE1-1,CITYA,WIDE2-1,CITYB would get to CITYB no matter where you are in
A or B or inbetween, but not involve ANY OTHER DIGI's than the shortest
path between them.  ETC...

Or CITYD,CITYC,CITYB,CITYA would let you go on a trip to CItyD, but your
packets would get back home to A no matter where you were inbetween.  When
close, then they would only go 1 hop.  Or if you wanted those at CityD to
anticipate your arrival, you might use the path CITYA,CITYB,CITYC,CITYD
and then your packets would all preceed you to CITYD until you got there,
but with fewer and fewer hops as you got close.

GATEWAY SCENARIO:  IN this scenario there is a special event on FREQB with
lots of FREQB digis that desires to get at least one (and only 1) copy of
each packet into the 144.39 system via a cross-channel gateway with the
callsign GATE.  The path of FREQB7-7,GATE,WIDE2-1 would make sure that a
packet got everywhere in the special event network but only one copy
always gets through the gateway at least one hop into the APRS network.
(such as the linear appalachan trail event or an underground multihop cave
event...

NO PREMPTIVE DIGIPEATING ON UIFLOOD and UITRACE.  The reason is, that it
would preclude any use of any kind of WIDEn-N AFTER going through a gate,
and that is one of the main reasons for doing it.

Remember, these 4 hop Premptive SPECIFIC paths are MUCH less load on the
network than a WIDE4-4 which can potentially cause 34 DUPES, compared to
only a maximum of 4 hops with the explicit preemptive hop approach.

Have I missed anything?

Bob, WB4APR




More information about the aprssig mailing list