aesop_modelisation (MIDAS)

Critical infrastructure modeling microservice — 5 PNAIV sectors, IS graph topology, H3 spatial tiling, cascade propagation engine.

MIDAS Internal Architecture

PNAIV SECTORS Communications Telecom, Data Centers IS1 IS2 IS3 Energie Electricite Nuclear Plants, Grid IS1 IS2 IS3 Energie Hydrocarbures Refineries, Pipelines IS1 IS2 Transports Road, Rail, Air, Maritime IS1 IS2 IS3 IS4 Eau Capture, Treatment, Distrib. IS1 IS2 Decomposition: PNAIV IS (n per) Nodes + Edges IS GRAPH MODEL IS: Reseau_Telecom_National N1 state: 200 N2 state: 200 N3 state: 200 w: 3.0 w: 2.5 w: 1.0 IS: Reseau_Electrique_RTE N4 state: 200 N5 state: 200 N6 state: 200 w: 4.0 w: 2.0 cross-IS dep Node (facility) Edge (connection) Cross-IS dependency operational_state: 200 = nominal capacity_weight (w): link capacity H3 SPATIAL LAYER N N N N H3 hex tiling (res 4-7) Altitude Layers GEO (35,786 km) MEO (2,000-35,786 km) LEO (160-2,000 km) THA (tropo high alt) HA (high altitude) BA (base altitude) SURFTER (surface terre) SURFMER (surface mer) VOLMER (colonne mer) FONDSMER (fonds marins) INJECTION & PROPAGATION Functional by IS/PNAIV type Geographic by H3 hex zone Manual node-by-node Cascade Propagation Engine Dashboard IEG_NOEUDS IEG_IS IEG_PNAIV IEP (cross-dep) DELTA indicators app_midas_api REST endpoints aesop_simulation OODA loop consumer

MIDAS Django Applications

App Purpose Key Models
app_midas_core Shared configuration, base classes, constants MidasConfig, OperationalStateChoices, PropagationProfile
app_midas_pnaiv 5 PNAIV critical infrastructure sectors Pnaiv, SecteurActivite (Communications, Energie Electricite, Energie Hydrocarbures, Transports, Eau)
app_midas_is Information System graphs: nodes, edges, cross-IS dependencies InformationSystem, Noeud, Arete, DependanceInterIS
app_midas_geo H3 hexagonal tiling and altitude layers for spatial indexing H3Cell, CoucheAltitude, NoeudGeoLocalisation
app_midas_propagation Cascade engine: BFS/weighted propagation across IS and cross-IS PropagationRun, PropagationStep, CascadeResult
app_midas_injection Three disruption injection modes: functional, geographic, manual InjectionScenario, InjectionPoint, InjectionMode
app_midas_dashboard Aggregated indicators and visualization for analysts IEG_Noeuds, IEG_IS, IEG_Pnaiv, IEP_CrossDependency, DeltaIndicator
app_midas_api REST API exposing graph state, propagation results, indicators Serializers, ViewSets (read-only + simulation triggers)

Architecture Concepts

Dual Storage Strategy

MIDAS uses two complementary storage engines on the same PostgreSQL instance:

  • Django ORM (relational) — Standard CRUD operations, admin interface, foreign keys for sector/IS/node hierarchy. Powers the dashboard and API serializers.
  • Apache AGE (graph) — Cypher-based traversal for propagation queries. Each IS is a labeled subgraph; nodes and edges are mirrored from ORM. Cross-IS dependencies are edges between subgraphs.

Writes go through Django ORM first, then sync to AGE via signals. Reads for propagation use AGE directly for O(E+V) traversal performance.

IS Graph Model

Each Information System (IS) is modeled as a directed weighted graph:

  • Nodes (Noeuds) — Physical or logical facilities: telecom towers, substations, refineries, water plants. Each has an operational_state (200=nominal, 300=degraded, 500=destroyed).
  • Edges (Aretes) — Connections between nodes with a capacity_weight representing flow capacity. Directed: a power line from plant to substation is not symmetric.
  • Cross-IS Dependencies — A data center (Communications) depends on a substation (Energie Electricite). These are modeled as inter-graph edges with their own dependency weight and criticality flag.

H3 Spatial Indexing

Every node is geolocated on an H3 hexagonal grid (resolutions 4-7) with an altitude layer assignment:

  • Hex tiling — Uber's H3 provides equal-area hexagons for geographic injection: "disable all nodes in these hexes" simulates area-effect disruptions (bombing, EMP, flood).
  • 10 altitude layers — From FONDSMER (seabed) through SURFTER (ground) up to GEO (geostationary orbit). Allows 3D modeling of satellite links, submarine cables, and aerial infrastructure.

Cascade Propagation Engine

The core analytical capability of MIDAS. When nodes are disrupted (via injection), the engine propagates effects:

  • Intra-IS propagation — BFS/weighted walk within a single IS graph. A destroyed substation degrades all downstream nodes proportionally to edge weights.
  • Inter-IS propagation — Crosses dependency edges. A power outage cascades into telecom (no power to towers), then into water (no SCADA communications).
  • Output indicators — IEG_NOEUDS (per-node), IEG_IS (per-IS aggregate), IEG_PNAIV (per-sector), IEP (cross-dependency exposure). Delta indicators track change from baseline.

Operational Modes

Standalone Mode (MIDAS for civilian planners) — MIDAS can operate independently as a critical infrastructure resilience analysis tool. Civilian agencies (SGDSN, ANSSI, prefectures) use it to model national infrastructure, simulate disruption scenarios, and assess cascading failure risks. In this mode, injection is manual or scenario-based, and the dashboard provides direct analytical outputs without feeding into a military simulation loop.
AESOP Substrate Mode (military integration) — When deployed as part of the AESOP trilogy, MIDAS serves as the infrastructure modeling substrate for aesop_simulation. The simulation's OODA loop engine consumes MIDAS graph state and propagation results via app_midas_api. Strike planning in the simulation triggers injection scenarios in MIDAS, and the resulting cascade indicators feed back into operational decision support. Intelligence from aesop_intell can also enrich the MIDAS model by updating node states based on OSINT signals.