Proportional Pathing and Decayed Beaconing 10 Nov 08 ----------------------------------------------------------------------- WB4APR 10 Nov 08 updated to include trackers 23 Mar 06 Original draft Proportional Pathing was introduced in 2006 as an additional improvement on the original APRS decaying algorithm. It was immediately implemented in all digipeaters under the New-N Paradigm and also in the new Kenwood D710 radio. It is recommended for ALL future APRS applications. With PROPORTIONAL PATHING, packets go frequently locally and direct, and less frequently at more and more hops. For digipeaters, this significantly cut the load on the network. Now digis ID once every 10 minutes local direct, once every 30 minutes via 1 hop, and once an hour via 2 hops. THe local 10 minute updates do not QRM with any users, since the digi will not TX if the channel is busy. For Mobiles, Proportional Pathing provides good local tracking continuity and presence, but overall, much less load on the APRS network. Also, it works for both routine operations and special events without any setting changes by the operator! Operator setting errors is one of the biggest problems with paths on the APRS network. Here is what Proportional Pathing does for a mobile: - Every minute the packet goes DIRECT path (no digi) - Every 2nd minute it goes via WIDE1-1 path - Every 4th minute it goes via WIDE1-1,WIDE2-1 - Every 8th minute it goes via WIDE1-1,WIDE2-2 (only if 3 hops is OK) The good news is that any existing KPC-3+ Digipeater OR tracker can do this already. See: www.aprs.org/txt/P-Pathing-kpc3.txt Proportional Pathing of digis and mobiles drastically reduces channel load and should be included in all new designs and firmware upgrades. The beauty of Proportional pathing is that it does not need to be tweaked by the user as he changes from routine to special event. or goes from area to area. Always his packets are higher rate close in and lower QRM at a distance. Tackers can pre-program these settings and never worry again. Of course some simple on/off switch options should be considered. One is the HOPS 2/3 switch, and the other is the RATE 2/3 switch. The HOP switch determines how far out the longest path goes. In the above example it is set to 2. If it was "3" then every 8th minute a packet would go via WIDE1-1,WIDE2-2 for a total of 3 hops. The RATE switch determines how rapidly the QRM falls off with distance. In the above examples, the rate is "/2" meaning the rate goes down by two at each additional hop. But if the RATE switch is set to "/3" then the QRM is further reduced since at each additional HOP tier the rate is 1/3rd the rate. For example: - Every minute his packet goes DIRECT path (no digi) - Every 3rd minute it also goes via WIDE1-1 path - Every 9th minute it also goes via WIDE1-1,WIDE2-1 - Every 27th minute it also goes via WIDE1-1,WIDE2-2 The beauty of Proportional Pathing is simply that other than the on/off switch options above, it does not need precise tweaking for most situations. Yet works at either extreme, local and multi-hop. OPEN ROAD: The local 1 minute update rate is also ideal on the open road out of range of all digis. With joint-crossing speeds of 140 MPH on the open road, you need a 1 minute rate if you want any chance of hearing another local APRS mobile running Voice Alert. DECAYED BEACONING: It should be noted that APRS was never designed for fixed rates. Because APRS recognizes the priority that new fresh data should have over old stale data! Too many APRS clones ignored this fundamental principal and operate at fixed rates that are either too HIGH and cause QRM, or are too low and provide little real-time continuity. The Decay algorithm was fundamental because only new-changed data was transmitted 'now', and that if the data did not change, then the time to the next transmission of that data would double. So, for positions for example, set to a 1 minute rate, if the mobile stopped for any reason, then the next packet would go out 2, then 4, then 8 then 16 and then 30 minutes later. As soon as the position changed (new data) then the rate was reset back to the selected rate (1 minute). Too bad so many APRS clones and follow-on devices simply used fixed rates for simplicity even when the data is not changing. This really added QRM to the network! THe decay algorithm is trivial to implement. Mobiles should simply compare new position data with the last transmitted data, and if unchanged, double the period to the next packet (to a max of 30 mins). RESULTS: The result is that for close in work or meeting someone or local events or whatever, the tracker gets good smooth high rate tracking (1 minute for example) and minimum QRM to anyone else. For the local area (each digi he can hit), he gets a comfortable 2 minute rate. For a region and those watching out 2 hops, he gets a 4 minute rate. And for observers a long distance away, he gets an 8 minute rate. And with decayed algorithm, if he stops, he also generates less QRM too. Yet if he operates his tracker at a special event, even without making any changes, he is also seen once a minute direct which is a good rate for an event. This resolves the tradeoff between too high a rate (QRM)and too low a rate (LATENCY, and LOW TRACK QUALITY)... PROPORTIONAL PATHING / Decayed Beaconing would cut the QRM from each mobile by a factor of 8 or more while moving, and 30 to 1 when stopped. And since mobiles are the largest load on the network, this would make a HUGE difference to the network while still providing good local tracking. HUMAN INTERFACE: The options for setting up a mobile APRS unit (tracker) in my opinion should simply be the selection of the HOP count and RATE divisor as introduced above. The basic rate is always 1 minute (if changing): HOP SWITCH: 2 or 3 RATE SWITCH: /2 or /3 This is just a continuation of the original APRS decay-algorithm which causes older packets to be transmitted less and less often and new data to be updated quickly. In the case with Proportional Pathing, it also makes NEAR data more important than FAR data as well as NEW data versus old duplicate data. Also, since the WIDE1-1,WIDEn-N should work everywhere, this one- size-fits-all appproach would make setting up a mobile tracker trivial, and help make the network immune to user set-up errors. This is part of the recommended APRS1.2 Addendum: http://www.aprs.org/aprs12.html de WB4APR, Bob