Monitoring & detection

Risk state machine

A risk state machine models hazard status as a set of discrete states with defined transitions between them, so that events are generated by changes of state rather than by individual readings.

Why it matters

Treating every reading as an event produces either an alert flood or a missed incident. Transitions are the meaningful unit: entering DANGER is news, remaining in DANGER for the next reading is not.

How it works

Readings drive the current state. Crossing from one state to another fires the side effects - creating an incident, raising an alert, changing reporting behaviour - while readings that leave the state unchanged do not.

Example

A node sitting at WARN for twenty minutes produces one incident on entry, not one per reading.

In Alfa HazardX

Incidents and alerts are created on transitions into DANGER and CRITICAL, and the threat timeline merges incidents and analyses into a single feed.