Link-Maintenance Sweep¶

Maintenance process — instantiates Spanning Connectivity Formation

Keeps an already-spanning network from quietly falling apart by detecting and repairing decaying links before enough of them fail to drop the network back below its connectivity threshold.

Link-Maintenance Sweep is the steady-state mechanism that defends connectivity already won. Every other sibling reaches toward a spanning network; this one keeps a formed one from decaying back out of existence. Links rot, expire, silt up, or lose authorization on their own, and the sweep runs periodically to catch that decay early and repair it while it is still cheap. Its defining insight is hysteresis: the effort to re-form a lost connection is far greater than the effort to keep it alive, so the sweep is triggered by a link trending toward failure, never by the failure itself. It is the one mechanism in the set that faces backward — protecting the threshold rather than crossing it.

Example¶

A rural county's villages are tied together by a road network that runs over aging timber bridges and culverts. No new connectivity is needed — but a single scoured-out culvert can sever a village, and rebuilding after a collapse costs many times a timely repair while stranding the community for a season. The sweep sends inspectors on a fixed cadence; each crossing is rated against a decay model of scour, load cycling, and rot, and the ones trending toward failure are repaired while still passable. A hysteresis monitor watches the gap between "degraded but usable" and "failed and impassable," triggering work in the first band so no crossing is ever allowed to reach the second. Through ordinary attrition the network stays spanning and no village drops off the map.

How it works¶

Model how each link decays. Give every link a decay signature — rate and failure mode — so degradation can be anticipated rather than discovered.
Sweep to detect the trend, not the break. Inspect or instrument against the decay model to catch links heading toward failure; detection is the front half of the job.
Repair in the degraded band. Act while the link still works, because a working-but-decaying link is far cheaper to fix than a failed one to rebuild.
Hold the hysteresis margin. The re-formation threshold sits above the break threshold; the sweep keeps every critical link on the safe side of that gap.

Tuning parameters¶

Inspection cadence — how often to sweep; faster catches decay earlier but costs more, so it is tuned to the link's decay rate.
Repair-trigger band — how degraded a link may get before action: early is expensive but safe, late is cheap per visit but risks a break.
Coverage — sweep every link, or only the cut-critical ones whose loss would actually fragment the network.
Decay-model refresh — how often to re-estimate decay rates as conditions shift (climate, rising load, ageing).
Hysteresis margin — how much buffer above the break threshold to hold; a bigger margin is safer but keeps more redundant capacity in service.

When it helps, and when it misleads¶

Its strength is preventing the slow, invisible loss of hard-won connectivity, and making explicit the asymmetry between cheap prevention and expensive rebuilding. Its failure mode is deferred maintenance: a decaying-but-working link looks fine, so the sweep is the easiest line to cut, and the damage stays hidden until a break cascades. This asymmetry — that crossing back below the threshold costs far more than staying above it — is exactly the hysteresis the monitor exists to respect.^[1] The classic misuse is running the sweep as a checkbox inspection that dutifully logs condition but never funds the repair, leaving an accurate decay model that everyone ignores. The discipline is to bind the sweep to a funded repair trigger and to prioritize the cut-critical links whose loss fragments the whole.

How it implements the components¶

Link-Maintenance Sweep fills the steady-state, decay-facing side of the archetype's machinery:

maintenance_decay_model — it models how each link degrades over time, the input that lets it act on a trend rather than a failure.
threshold_hysteresis_monitor — it tracks the gap between the break threshold and the higher re-formation threshold, triggering repair before the network can drop across it.

It does not build connectivity in the first place (candidate_activation_set — Link-Activation Program), watch overall reachability and coverage (Giant-Component Dashboard), or change the topology so a single loss no longer matters (Redundant Network Link and Cut-Set Hardening).

Instantiates: Spanning Connectivity Formation — it holds a formed network above its connectivity threshold against ongoing decay.
Consumes: Link-Activation Program hands off the freshly built network for the sweep to keep alive.
Sibling mechanisms: Link-Activation Program · Cut-Set Hardening · Redundant Network Link · Giant-Component Dashboard · Segmentation or Quarantine Switch

Notes¶

Maintenance is not hardening. This sweep keeps existing links alive; Cut-Set Hardening and Redundant Network Link change the topology so that any single link's loss matters less. The two are complements — a well-swept network can still be fragile if it spans through a single cut edge, and a hardened network still decays — but they operate on different things and should not be conflated in a maintenance budget.

References¶

[1] Hysteresis — the property that a system's path back to a prior state is not the mirror of the path that left it. Here, the density at which a broken network re-forms is higher than the density at which a formed one breaks, which is why timely maintenance is structurally cheaper than post-failure rebuilding. ↩