A relay network passes a message beyond one signal’s range by having each station repeat exactly what it received, acknowledge the station before it, and log the time. Run a test message at a fixed hour every day so a dead or sleeping station shows up as the point where the test stops, number every message so it cannot circulate forever in a ring, and expect the staffing cost to be the real limit.

A relay network is a set of stations, each within signalling range of its neighbours, that pass a message along from one end to the other.

Shapes Of A Network

  • Chain. Stations in a line. Simple, and one failed station breaks it.
  • Ring. Stations in a loop with signals able to go either way round. A ring survives any single break, as the telegraph article states for wire networks (Telegraph Network), but a message can circulate forever unless it is numbered.
  • Branch or star. One central station with spokes. Fast to the centre, and the centre is a single point of failure.

Staffing Is The Cost That Decides

A station that must always be ready needs someone watching it all the time. The watch-scheduling article’s table gives 4 to 5 people to keep one post watched round the clock (Watch and Patrol Scheduling).

Worked Example: A 60 Km Chain

Example inputs: 60 km between the two settlements, stations 10 km apart.

  • Stations between the ends: 60 / 10 - 1 = 5.
  • People to man five stations continuously at 4 to 5 per post: 5 x 4 = 20 to 5 x 5 = 25.
  • If only one settlement staffed it, a small community would give up most of its adults to a chain that carries almost no traffic.

Ways to cut the cost: split the stations between the two settlements; man the chain fully only at higher threat levels, as the Green, Yellow and Red postures in the watch-scheduling article do for local posts; or use runners for routine traffic (Runner Systems).

Rules For A Relay Station

  1. Repeat exactly. A station does not summarise, correct or add. If a message looks wrong, it repeats it and also reports the doubt.
  2. Acknowledge the sender before forwarding. An acknowledgement tells the sending station it was seen and stops it re-sending.
  3. Log the time and the message. Each station keeps a log (Log Keeping).
  4. Number every message. A serial number and the originating station mean a station that has already passed it on can drop a repeat. In a ring this prevents endless circulation.
  5. Forward in both directions only for alarms. A routine message goes one way.
  6. Never relay what fails authentication without telling the next station that it failed (Authentication and Security).

Method: Running And Testing A Chain

Step 1: Fix the stations by line of sight. Check each pair in daylight and at night.

Step 2: Agree the code book and the log format at all stations.

Step 3: Send a test every day at a fixed hour. The first station signals a harmless test; each passes it on. A station that has not heard the test within a few minutes flags the one before it.

Step 4: When the test stops, send a runner to the last station that passed it, not to the first.

Step 5: Review monthly. Count the tests that completed, the stations that missed, and the time from first signal to last.

Common Mistakes

  • Relaying without checking. One false message then reaches everyone.
  • No test. A chain untested for a season is a chain that does not work.
  • Single channel. If the stations have only a lamp, a bright day or a fog stops them.
  • No ownership. Stations between two settlements belong to nobody, and nobody keeps them.

What’s Next