Skip to content

Phase-Slip Recovery Protocol

Protocol — instantiates Decentralized Phase Locking

When a unit loses lock, a bounded re-acquisition maneuver that restores the useful phase relation without snapping back so hard it disturbs its neighbors.

Locks break. A shove, a stall, a burst of noise, and a unit slips out of its phase relation. Phase-Slip Recovery Protocol is the reactive maneuver that brings it back: it detects that the relation has left its band, switches from ordinary coupling into a rate-limited re-acquisition, and steers the unit back to the functional relation — not to zero-error perfection. Its defining discipline is restraint in two directions at once. It aims for the useful band rather than a perfect lock, and it caps how fast it corrects, because a unit that yanks itself violently back into position transmits that yank to everyone coupled to it — turning one slip into a wave. Recovery that destabilizes its neighbors is not recovery.

Example

A convoy of autonomous trucks runs with intentionally staggered brake-and-accelerate cycles, so they never all decelerate at the same instant and set off a shockwave down the line. A hard stop knocks the third truck out of its slot — a phase slip; its offset has drifted past the safe band. The recovery protocol notices, and rather than jamming the truck back into exact position (which would throw a jolt onto the truck behind it), it retunes its cadence over several cycles, easing back into the ±X-second window. It targets the functional stagger — safe non-overlap of braking — not a perfect phase, and it rate-limits its own correction so the recovery itself stays smooth. Once inside the band, it hands control back to normal coupling and logs the slip.

How it works

The protocol is a small state machine distinct from steady-state coupling:

  • Detect loss of lock. A slip is declared when the phase relation leaves its band or a slip is counted — not on ordinary jitter.
  • Switch to a rate-limited recovery mode. Normal small nudges are replaced by a bounded, retuned correction sized to re-acquire without shocking neighbors.
  • Converge to the functional band, then hand back. It restores the useful relation — an offset, a stagger, a bounded error — and returns to normal coupling, logging the event.

Tuning parameters

  • Lock-loss threshold — how far out of band triggers "slipped." Tight recovers quickly but false-triggers on noise; loose tolerates jitter but reacts late.
  • Re-acquisition rate limit — the cap on correction speed. Fast recovers sooner but risks shocking neighbors; slow is gentle but leaves the relation broken longer.
  • Recovery-target width — how tight a relation to restore. Restoring the functional band is cheap and stable; chasing perfect lock costs effort and risks overshoot.
  • Hysteresis — the gap between declaring "slipped" and declaring "recovered," to stop flapping at the boundary.
  • Fallback — what happens if re-lock fails after N cycles: drop out, enter a safe mode, or escalate to retuning.

When it helps, and when it misleads

Its strength is robustness: a population whose units re-lock gracefully after disturbance beats one that locks more tightly but shatters when hit. It is what makes an emergent lock survive the ordinary shocks of operation.

It misleads when recovery is too aggressive or too frequent. Chasing a perfect re-lock too hard transmits the original disturbance onward — the recovery becomes the next unit's shock, exactly the wave the stagger was meant to prevent.[1] And repeated slips are usually a signal that the underlying tuning is wrong, not that more recovery is needed. The classic misuse is leaning on recovery to paper over chronic detuning that should be fixed at its source. The discipline is to target the functional band rather than zero error, rate-limit the correction, and escalate to diagnosis when slips recur.

How it implements the components

  • local_adaptation_and_retuning_rule — recovery is a bounded retuning: on a slip, the unit switches to a rate-limited correction distinct from its ordinary nudges, adapting to re-acquire the relation.
  • functional_coherence_target — it re-acquires the functional relation — a bounded band, an offset, a safe stagger — not zero-lag perfection; that target is what defines "recovered."

It does not deliberately inject faults to test recovery — that is Perturb-and-Relock Drill — nor map the capture basin it must steer back into, which is Weak-Coupling Ramp Trial; it consumes that basin and reacts to real slips.

References

[1] In power systems a generator that falls out of step — a loss of synchronism or pole slip — is brought back by a controlled re-synchronization or a deliberate trip, precisely because an abrupt, unmanaged re-lock injects a large transient into the rest of the grid. The same logic drives the rate limit here. ↩