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: Networking Notes from N5EG (TEXNET) (4 of 4) Message-ID: <372@timeinc.UUCP> Date: Wed, 31-Jul-85 15:59:09 EDT Article-I.D.: timeinc.372 Posted: Wed Jul 31 15:59:09 1985 Date-Received: Sat, 3-Aug-85 02:42:17 EDT Reply-To: dwight@timeinc.UUCP (Dwight Ernest) Distribution: na Organization: Time, Inc. - New York Lines: 127 [ Note: This series of articles was found on Compuserve and downloaded from HAMNET there on 21 July 1985 by Dwight Ernest KA2CNN 70210,523. ] An Introduction to Networks part 4 by T.C. McDermott, N5EG networks SIG, TPRS This article describes some requirements for network node hardware. One of the key concepts is the idea of modular implementation - to allow for changes in the way a network is designed. The network that has been evolving from the description in the 3 previous articles can be implemented in a backbone type of network. In this network there are two channels that are accesible at any node, the high-speed inter-node communication channel, and the 2-meter AX.25 channel. Thus the description of the hardware assumes that there are two synchronous channels per node. Each node has the capability to move traffic along between nodes (all on the high-speed channel) and also to drop and add traffic to the high-speed channel from the low-speed channel. One of the first assumptions is that the high-speed channel will operate at 9600 bits-per-sec (b/s). For a number of reasons (the formost of which is availability) the K9NG type of FSK modem on the 220 Mhz. band will be utilized. There are some performance advantages to be obtained with PSK, but the slow carrier-recovery loops that are normally used are not always compatible with the fast T/R-switching times needed in packet service. The low-speed channel will compatible with AX.25 TNC's, and thus is a 1200 b/s channel on the two meter band. We have thus partitioned the node into three major elements: 1. High-speed channel hardware. This includes 220 MHZ. radio, antennas, power splitter, 9600 b/s FSK modem (K9NG). 2. Low-speed channel hardware. This includes 2-M radio, antenna, and 1200 b/s AFSK modem. 3. Node-control computer. This includes 2 synchronous interfaces, T/R control circuits, RAM, ROM, and fail-safe sequence decoder. A simplification of the high-speed radio circuitry is to have only one 220 Mhz. transmitter and one 220 Mhz. receiver per node. Two high-gain directional antennas are used, with a 3-dB. power splitter near the antennas. Thus communications with the northerly- and southerly- nodes is possible with one radio. Extension of this concept to three or four nodes is possible but the RF losses start to get high. The design of the high-speed portion (the backbone) allows access to the node only by other nodes. This is done to eliminate direct channel contention between users and the inter-node communication. The users of the network do not, and indeed CANNOT communicate with it on 220 Mhz. The protocol on 220 Mhz. (in this implementation) is TEXNET-IP, which IS NOT compatible with AX.25 (and indeed, for the reasons expressed in parts 1-3, should not be compatible with AX.25). The 2-Meter section can use a commercially-available FM radio, and AFSK modem design. Each of the radio sections contains logic-level interfaces to the node control unit. This is done to facilitate the changeout of the node control processor should a new design or network protocol standard become available. It is anticipated that resolution of the various trade-offs involved in the implementation of a network will take several years to occur. In order to solve a pressing need within TPRS, the need for long- haul communication, we will go ahead and implement TEXNET with an eye towards changes and evolving standards. The node control computer consists of a Z80-SIO chip (which has two synchronous HDLC-type serial channels), a Z80 (4 Mhz.) microprocessor, 16K of ROM, 32K or RAM, some timer circuitry to develop the 1200-hz, 9600-hz, and 1200 x 32 = 38400 hz. clocks. The node also has two time-out timers to prevent transmitter lock-up on the 2-m and 220 Mhz. units, contains NRZ/NRZI coder, decoder, and clock-recovery circuits for the 1200 b/s channel. It also contains a special state machine that listens to the 220 Mhz. channel and clock. This circuit recognizes a special sequence (that obeys the HDLC coding rules) and interprets the reception of this sequence as a over-riding command that uses hardware to re-boot the node processor. Each node contains a unique code in its state machine. The code is chosen to be sufficiently long that the mean time to false is 6*10^7 years (assuming random data). A custom circuit board will be constructed to contain this controller. It may have been possible to modify one of the Xerox-820 boards, but it was felt that the changes required would reduce the reliability of the resultant assembly too much. The parts cost of byte-wide RAMs and ROMs has dropped recently, and these devices will should allow construction of the entire controller for slightly less than the price of just the 820 board when purchased surplus. The controller will be constructed mostly of CMOS circuitry, and will be powered at +5 V through a series regulator powered from +12 VDC. This will allow a single +12 V supply. The node will contain a gel-cel battery and a charger circuit. Thus the entire node will have something approaching un-interruptable power, while still having an acceptable power supply cost. The node controller card will be connecterized at the logic- level interfaces to the radio circuits. In the event a new controller design emerges, then upgrading of the node can be as simple as replacing the card. One of the objectives of this network design is to keep the cost of any node low. Our goal was $500.00. We anticipate that those groups who wish to joint TEXNET will assemble, place, and maintain their node, with the assitance of TPRS. We thus would release the design of the node to those groups, and perhaps sell or supply the circuit boards needed, as well as the software for the controller. In our design, each node will have the same software, except for routing tables. A first crack at the routing problem can be attempted with static routing tables, whcih will be in ROM, and different at every site. The subjects of routing, and other network topics will be discussed in part 5 of this series. -- ----------------------------------------------------------------------------- --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." -----------------------------------------------------------------------------