Skip to content

Adaptive TX power for access repeaters - limit flood repeats to the local cell #3875

Open

Description

@usrflo

Problem

By design, every repeater rebroadcasts every incoming flood packet at maximum TX power. In dense urban deployments each repetition is heard - and blocks the channel via CAD - far across the city: overlapping ranges merge into one congested collision domain. Two measures exist to reduce this load: (1) reduce the number of repetitions of each packet, and (2) limit the reach of each repetition.

This proposal targets the incoming flood packets. For an access repeater flood packets like external node adverts or channel messages need to reach its own companions only - not the whole world.

Prior considerations

  • Flood suppression (Feature: redundancy-aware flood suppression for traffic reduction (simple_repeater) #3019) limits the number of repetitions where neighbours are redundant - but it is double-edged: the marginal companion attached solely and quietly to the suppressing repeater, or a distant repeater relying on it, may lose everything. The attempt to monitor surrounding nodes to create a reachability graph works but it's complicated and may fail in edge scenarios.
  • Cutting the reach of repetitions instead: a locally throttled repeater with a dynamically lowered TX power still serves its cell, while its repetitions stop blocking the city. Collision domains shrink and transmissions can run concurrently per cell.

Proposal: How to limit the local repeater's reach

Introduce an optional "access repeater" behaviour:

  • Refer to Inbound floods only. Reduced TX power applies only to flood packets arriving from outside the cell — those already carried by at least one other repeater. Floods that reach the repeater as the first hop (from its own clients and immediate neighbourhood) are still rebroadcast at max TX, and so are direct-route packets (messages with a defined path, where transport responsibility demands max TX). An admin's own traffic gets out to the wide mesh exactly as before.
  • The cell and its usage are defined by self-registered clients. Anyone holding the guest password of the repeater can log in and register as an active client of the cell — one prompt ("Register as active client? y/n"), the app sends its public key, no manual key handling.
  • The maximum power for sending into the cell is measured: the required TX power of the repeater focuses on the weakest registered client, clamped between configurable min/max. Starts at max power, only escalates on evidence; silent members age out; admin can list / kick registered clients.
  • No repeater topology knowledge required. The registered companions indirectly define the range the repeater needs to cover — it is not necessary to know every node in the mesh.
  • The uplink is anchored. The reach needed for the important neighbour repeaters is included in the measured level, so the repeater reliably gets its own - and its cell's - traffic in and out to the world.
  • This answers the suppression critique directly: the weakest registered and required node defines the floor while nothing is suppressed.
  • Goals reached: a more stable mesh and a reduced battery consumption on the access repeater.
image image

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions