Bulletins from the Pacific Packet Radio Society - page 059

specify a node by its individual address, by some common group ID, by class of service it can provide, and (by default) as a member of the global network for true broadcast capability.

Now to the response to comments:

"you must check into an NCS to connect ... 11 - Connecting any stations together requires some form of handshaking. This protocol, for the overhead of control packets at sign-on, will prevent much traffic that would go to stations which will not happen to be on-the-air at the time of connection attempt. Remember, there is no "Network Controller", just a friendly station who wants to help new stations determine who's on channel.

"the ability to communicate after the loss of a net controller..." - see 4) above.

"satellite's footprint complicates both addressing and handoff.." - Both addressing and re-addressing (handing off) are fully automatic, and the ground links will have access to all information as to location of a node at any instant with a single query to address servers connected to links either integrally or via the local channel. No global directories are required at a central site for a satellite inter-networking facility to operate.

" at HF, the propagation.. question the feasibility of net controlling at all.." - Once again, there is no Net Control. After receiving an address from some station, higher-level protocol within the node would be free to keep that address after propagation takes its toll on the paths involved. Note that this is a perversion caused by trying to use a local area network protocol in an extended network environment. It might be prudent to look at the entire problem before trying to fit existing answers.

My final comment is on the statement in the Internet Sublayer description, "..that would contribute to additional overhead which is universally frowned upon". - In designing the most efficient system without considering the level of throughput required, many simpler schemes which are not as efficient as more sophisticated (read "complex") techniques are thrown away. These alternatives should always be considered. Also an indirect impact of a piece of a protocol on total throughput of the channel is possible, by requiring more complicated mechanisms to use the "simple" service. Therefore proving a part of a protocol more efficient (e.g. "dynamic" vs "static call-sign" packet efficiency) does not guarantee the best solution. Always feel free to use multiple solutions to the same problem at different sites, with commonality guaranteed at the gateways. Why should we be different from the vendors of local area nets? (They're still experimenting too!)

73, Den KD2S

Click for Original Graphic Image of this page.