Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.2 9/18/84; site timeinc.UUCP Path: utzoo!decvax!decwrl!greipa!pesnta!phri!timeinc!dwight From: dwight@timeinc.UUCP (Dwight Ernest) Newsgroups: net.ham-radio.packet Subject: Thoughts on Packet's Future from MAPRC (2 of 2) (LONG) Message-ID: <377@timeinc.UUCP> Date: Wed, 31-Jul-85 17:06:35 EDT Article-I.D.: timeinc.377 Posted: Wed Jul 31 17:06:35 1985 Date-Received: Sat, 3-Aug-85 02:43:32 EDT Reply-To: dwight@timeinc.UUCP (Dwight Ernest) Distribution: na Organization: Time, Inc. - New York Lines: 224 SOME THOUGHTS ON THE PRESENT AND FUTURE OF PACKET RADIO IN THE MID-ATLANTIC AREA Tom Clark W3IWI 27 May 1985 A. Introduction: There has been a lot of discussion the past few months on how we should develop our packet radio networks in the future and here are some of my ideas. First let us begin by listing the a priori knowledge that we have: (1) Packet radio is experiencing an explosive growth. In each of our local areas in MAPRC (Balto/Wash, Philly/Trenton/SNJ, Harrisburg/York, etc.) we are seeing several new users pop up each week. The recent entry of Heath and Kantronics into the commercial market (complete with unprecedented ad campaigns), plus TAPR's TNC2, plus the series of articles that are to appear in QST this summer will all act like a magnet. In May alone, nearly 100 users have logged into the 'IWI BBS at least once. (2) Our present networks DO work. The BBS's are serving a communications need for non-real-time electronic mail (EM) between individuals, as a focal-point for public discussions ("electronic meetings") and as a way for users to get help. They are serving as our publishing channel (why spend $0.22 to mail this to you when I can post it by radio?) so that the MAPRC confederation hasn't felt the need to start a "paper" newsletter. Assuming that such communications must extend past the coverage area of any BBS and it's associated local area network (LAN), these communications functions REQUIRE linking. (3) Individual users want to have real-time links available to them too, whether for "conventional" QSO's or for computer-to-computer connections. Some users must use a part of the linking network just to gain access to the local BBS (e.g. the Virginia folks to use 'IWI or '3Q BBS or APR-5 HF Gateway). A growing number of users from further away want to use the linking "trunks" to access BBS's outside their LAN (the BBS DXing syndrome!). Some of the more technical types want to use the links to test advanced networking concepts before they are made "public". (4) The "community" resources which we all enjoy were, for the most part, installed by individuals and their ultimate control lies with PEOPLE. There has been precious little community (read that to mean you all) resources (read that to mean money, hardware, digipeater sites, etc.) applied to develop the networks. (5) Technology is rapidly improving -- K9NG has developed 9600 baud modems which will be available soon. TNC2 is already a reality, albeit still in a "beta" testing mode. AX.25 Rev.2 code which will fix some "bugs" is near at hand (it is already available in TNC2 and KA9Q X820 code). Rudimentary Level 3 (networking) software is running between the BBS's and certainly will be available for more general use within a year. W0RLI and his network of linked BBS's continues to enhance their capabilities daily. KE3Z has developed dual-ported digipeater code. (6) One of our most valuable resources are the frequencies we use. Most of the activity is in the 145.01-145.09 range since radio hardware is easy to obtain. This range is being established nation-wide for packet use, but in the TMARC area, we only have coordinated three channels (.01,.03,.05). 220 is not a panacea for curing all future ills for two reasons -- we may very well lose the band in the next 1 (3,5,??) years to commercial interests, and radio hardware is not readily available. Only one channel has been coordinated on 441.0 (where hardware is more readily available), and hardware is hard to come by on 1270. Because of all the above, our skeleton system-level resources are being taxed to the limit. Users thrash with BBS activity on 145.01. People still continue sending Beacons announcing that they aren't home but their computer is! New groups want to help by putting up more digipeaters and BBS's to serve their LAN needs. Is this problem soluble? I think it is, and here are some of my ideas on HOW. B. The User and the LAN side of things: Central to the problem AND the solution is the user and his LAN. My prediction for the future is that packet radio will evolve towards being a non-real-time communications service. Industry has already demonstrated that electronic mail (EM) is the only practical way to tie together diverse groups that are geographically "spread out". I want to communicate when it is convenient for ME and I don't want to have to worry about the other person's schedule. The problem is compounded if multiple people need to participate in a "meeting" -- N people either have N^2 schedule conflicts or they have N^2-N separate 2-way discussions -- in either case progress is retarded. I see the path towards orderly development involving the concepts of cellular radio. We have already seen that about 30-50 active users is about all that a BBS can accommodate without becoming a full-time job for the SYSOP and without having the individuals "thrash" for BBS access time (assuming a normal single-user BBS configuration like we have now). [ It is interesting to note that FM repeater groups have discovered the same thing. The "super" repeaters with hundreds of users are able to support only brief call-and-answer communication and usually "spin-off" a new repeater for every 100 members or so. ] I envision that we will see local-area "cellular" BBS's springing up to serve LAN needs. Most, but not all of these will provide their user access on 2M (at 1200 baud) since equipment is cheap and readily available. These "cellular" BBSs will be QRP since they only need to provide direct access to a small physical area. They probably do not need to have digipeaters associated with them unless terrain and similar local conditions dictate. Adjacent "cells" would have coordinated frequencies so that they don't hear each other. A theorem in topology states that ANY map can be "painted" in four colors, with no adjacent areas having matching colors; therefore given a perfect world, four frequencies should be able to have minimal thrashing given QRP LAN coverage by all stations. Given my premise that 2M is the LAN frequency of choice, and the present 145.01-.09 band, these four channels are .03, .05, .07 and .09 (if we were to choose to enter the 15 vs. 20 khz repeater war, we could stretch this to provide one more channel by adopting 15 khz spacing). C. Linking the LAN's: Nobody wants to "talk" only with people in his own LAN -- that's what networking is all about. At present EASTNET exists only on 145.01 and thrashing is occurring because LAN functions go on at the same time as legitimate network activities. INDIVIDUAL USERS AND NETWORKS "TRUNKS" CANNOT COEXIST!! In the evolution of our system, the network should serve to interconnect the LANs, and the BBS's (or other gateway stations) provide the user interface into the network. INDIVIDUAL USERS SHOULD NOT ACCESS THE NETWORK "TRUNKS" DIRECTLY. Our present networks are fragile and easily broken. If WB2RVX points his beam north to access WA2SNA, then access thru Mike from APR-6 is marginal; and Mike is an experimenter at heart -- sometimes his station is doing other things. A single-point failure at a key site causes us to lose everything. The present bastardized network using dumb digipeaters requires end-to-end acknowledgements. Let us assume that a station has a 90% probability that a given packet will get to the next station, and a 90% probability of receiving the ack back. Then .9*.9 of his packets = 81% of his packets will make it thru (with ack) on the first try, and he will have to retry once out of 5 transmissions. This presents no serious problem other than the fact that 19% of the channel time is being wasted. However if 7 hops were needed, then the thruput goes down to 0.9^14 = 22.9% and only one packet out of 5 makes it the first time! The result -- the channel is clogged and nothing works. The Level 3 development work is intended to remove the need for end-to-end acknowledgement. The BBS's right now do this Level 3 function now in their mail forwarding -- IWI does not attempt to connect with W0RLI to get stuff to Boston, instead IWI passes the buck by sending it to WB2MNF, who then worries about the next hop up the coast. The problem is aggravated if one of the key relay stations is situated in the midst of heavy LAN activity. Typical local users will be running small antennas and simply not hear the remote stations at all; the result is that LAN users on link frequencies kill the link! INDIVIDUAL USERS SHOULD NOT BE ON LINKING FREQUENCIES! At present, all we have is .01 (and HF) for linking. The same frequency is used by users and network gateways. Users also need to access the network directly since they do not have direct access to the gateways. The solution seems to me to be to build a second, redundant network in parallel on different frequencies. This second network should run the highest possible speeds consistent with reliability and affordability. Right now, the winner seems to me to be the K9NG 9600 baud design. Hopefully boards will be available soon. To go along with these modems, we need to begging implementing multi- (or at least dual-) ported digipeaters; the KE3Z design based on the venerable X820 board seems to be the logical choice. And we need to establish a source of good, reliable radios and antennas. It appears that 220 is the best choice, even considering that 220 is new "turf" for many of us, and that hardware is harder to find, and that the FCC might choose to cave in to the commercial interests and pull the band out from under us. If not 220, then the next place we have to go is 1270, and that will probably be even a tougher nut to crack since even less hardware is now available. The 2M links on .01 should be maintained. For now it is all we have. When the higher speed links are in place, .01 can serve as an independent backup. And .01 can be the frequency for individual users to gain access to the longer-haul links for their own purposes. D. A Plague on BBS DXing!: As a special case of the problem just discussed, I want to cite a growing cancer in our midst -- BBS DXing!! The past couple of months have seen a large number of new users come on. The links have been improving with the addition of WA2SNA-2 in NNJ. Here I have seen a growing number of new users who seem to be unaware of the fragility of EASTNET. The will call in to IWI with 5-6 hop paths that look like (the call xxx is suppressed to protect the actual offenders from Bronx cheers!): xxx <=> WA1IXU <=> KG1O-9 <=> WA2SNA-2 <=> WB2RVX <=> WB4APR-6 <=> W3IWI Somehow, their one-way connect request packet arrives at IWI and with the AX.25 handshaking, the BBS send a connect ack and assumes that we are connected. It then tries to log the person on, and tries again, and again, and again until the BBS's timers say "Sorry Charlie" and hang up. Then xxx, having perhaps seen a packet or two coming thru from the BBS sez "Geez, this is nifty. I think I'll try again" and starts the whole cycle over again. In a 13 day sampling period in May, of 273 connections logged on the IWI BBS on 145.01, a total of 96 (=35%) ended with timeouts. INDIVIDUAL USERS SHOULD NOT ACCESS BBS'S THRU MORE THAN 2-3 DIGIPEATERS!!! BBS's are there to serve LAN and network gateway functions -- BBS DXING SHOULD BE BANNED!! I am considering adopting the anti-social solution to this problem by putting those folks who show up in the "timeout" list onto the "banned user" list and invite comments on the acceptability of this approach. E. Getting it all together: The fragile system we now have was built by individuals as an experiment. It works! It paves the way for the future. The individuals cannot be expected to continue to provide ALL the hardware as a gratuity just 'cuz they are nice guys. The new wave of users who want the services will have to begin carrying their share of the load. The LAN BBS's and local coverage digipeaters are the logical province of local clubs and groups. The shared resource, to which all must contribute is the interconnecting network. Parts of the network are going to require WORK!!! New sites for key linking stations have to be acquired AND MAINTAINED. Special hardware -- high speed modems, multi-port digipeaters, radios, etc. -- all have to be built and paid for. And the whole network has to be coordinated. I envision the Mid- Atlantic Packet Radio Council -- MAPRC -- serving that coordination role. It remains to be seen if that role includes only technical coordination or if it also will include the financial/managerial functions too. I ask MAPRC to consider these ideas and begin forging packet radio's future destiny NOW. If we don't act now, then I feel that we are in danger of having uncontrolled chaos in the future. -- ----------------------------------------------------------------------------- --Dwight Ernest KA2CNN \ Usenet:...vax135!timeinc!dwight Time Inc. Edit./Prod. Tech. Grp., New York City Voice: (212) 554-5061 \ Compuserve: 70210,523 Telemail: DERNEST/TIMECOMDIV/TIMEINC \ MCI: DERNEST "The opinions expressed above are those of the writer and do not necessarily reflect the opinions of Time Incorporated." -----------------------------------------------------------------------------