Loading...
How the platform analyzes individual countries — connectivity, resilience, failure impact, and strategic network position.
Country analysis operates on the country-level topology graphand produces four major sections: Connectivity (infrastructure inventory), Resilience (how well-connected the country is), Impact (what happens when things fail), and Position(the country's strategic role in the global network).
For every country reachable from this one: shortest hop count, path count, whether it's directly connected, and the precision of the path count (exact or approximate).
A BFS flood-fill from the source country discovers all reachable countries and their distances. Then, for a selected subset of destinations, exact bounded BFS path enumeration runs to count real paths. The selection prioritizes:
For all other destinations, path count is approximated: for direct neighbors, it uses the cable count on that edge (capped at 5); for indirect, it returns 1.
Path precision is mixed. Only ~22 destinations get exact counts; the rest are approximations. The approximation for indirect countries (always 1) is a rough lower bound. The result includes a pathPrecision field so consumers know which values are exact.
Full exact enumeration for all destinations, or a smarter approximation using edge multiplicity along the shortest path. Parallel BFS computation.
For each other country in the network, what happens if it goes offline: how many of this country's reachable destinations are lost, and how many paths are affected.
For every country in the baseline reachable set: simulate blocking that country, then BFS from the source to see which destinations are still reachable. A weighted transit dependency score is computed:
Only countries whose loss actually affects the source country are included.
The scoring formula is an arbitrary weighted heuristic. The weight choices (100/10/5/2) strongly prioritize reachable country count over other factors. Path counts for non-exact-precision destinations use approximated values in the score.
Measures how concentrated connectivity is across three dimensions:
Uses the Herfindahl-Hirschman Index (HHI): sum of squared market shares. For each dimension, the "share" is the fraction of reachability (or cable count) attributable to each entity. HHI ranges from near 0 (evenly distributed) to 1.0 (fully concentrated on one entity). Reports top 5 entities per dimension with their shares.
Gateway country share is based on first-hop analysis from exact-precision paths only. For countries where most destinations have approximate precision, the gateway concentration may not reflect the full picture.
A single number summarizing how well-connected the country is.
The formula is:
It combines the fraction of globally reachable countries the country can reach (0–100 scale) with the total number of alternative paths across all destinations.
The score mixes two different scales (percentage + raw count) and is not normalized to a fixed range. A well-connected country can score well above 100, making cross-country comparison unintuitive.
Normalize to a 0–100 scale using percentile ranking across all countries, or use a multi-dimensional scoring system with separate axes.
Simulates failure of each individual cable and each individual landing point, measuring: which direct connections are lost, whether the country remains connected to the global network.
Cable failure: For each cable, check which other countries lose their direct connection (no other cable connects them). Check global network reachability with this cable blocked. LP failure: For each LP, determine which cable-country edges are broken. An edge is broken only if the LP is the sole LP for that cable in that country. If a cable has multiple LPs in the same country, losing one does not sever the cable-country connection.
Only N+1 (single-asset) failures. Does not simulate cascading failures or multi-asset outages. The LP failure model assumes other LPs in the same country provide implicit redundancy, which may not hold if they share physical routes.
The most damaging pairs of failures (cable+cable, LP+LP, cable+LP) that isolate the country from the most destinations.
Uses a bounded candidate search:
Isolation bonus: combinations that fully disconnect the country get a +1000 score boost.
Only tests combinations from the top 5 single-failure candidatesin each category (25 cable pairs, 25 LP pairs, 25 mixed pairs = 75 total). A combination of two moderately-impactful cables might be worse than a combination involving a top-5 cable, but it won't be found.
Expand candidate pool, or use graph-theoretic approaches (min-cut pairs) to find optimal failure combinations without exhaustive search.
The country's strategic role in the global network, classified as:
Computes percentile rankings of the country's degree, reachable country count, and top gateway share against all countries. Uses threshold-based rules: e.g., Gateway requires degree ≥ 75th percentile AND reachable ≥ 75th percentile AND top gateway share ≥ 75th percentile. Bridge coverage is the fraction of bridge edges the country is an endpoint of.
Classification thresholds are heuristic percentile cutoffs. Countries near threshold boundaries may flip between roles with small data changes. The "Mixed" category is a catch-all that provides limited insight.