Per-node sensor capabilities for a 20' ISO container reference design (Tethra Laulima HG-C1 class — 120 kWp PV, 264 kWh battery, 9m telescoping mast, autonomous drone dock). Ranges are conservative estimates from public defense and academic sources. The full sensor electronics stack fits in 1-3 rack units inside the container at ~50-150W continuous draw — <1% of daily PV harvest.
This atlas is a concept demonstration, not a precise range model. Coverage is rendered as uniform circles for visualization. Real-world performance varies significantly with terrain (line-of-sight obstructions, mast height), atmospheric conditions (ducting, rain attenuation, fog), sea state, sea spray, antenna obstruction, target signature (EIRP, RCS, IR signature), and sensor fusion benefits across the network. Treat the displayed envelopes as architectural illustrations of the concept's shape and scale, not as commitments about specific link budgets.
Cooperative
AIS — 20 NM ship-to-ship. VHF Channel 87B/88B (161.975/162.025 MHz). Cooperative civilian protocol; spoofable but mature TDOA/SAR cross-correlation defeats most spoofs. Reference hardware: em-trak R100 or dAISy Pro-class receiver (book-sized, ~$200-2K) with marine VHF whip on mast head. Range scales with antenna height — 9m mast yields ~20-25 NM under nominal refraction. Source: ITU-R M.1371; document §AIS
ADS-B — 250 NM. 1090ES Mode-S broadcast from cooperative aircraft. Range scales with elevated antenna and is essentially horizon-limited at altitude. Reference hardware: FlightAware Pro Stick / RadarBox FlightStick (USB-dongle scale, ~$300) with vertical or log-periodic antenna under 1m tall. From a 9m mast to FL300 traffic, line-of-sight reaches ~220 NM — the modeled 250 NM accounts for higher-altitude traffic. Source: ICAO Annex 10
Passive RF SIGINT (broadband SDR)
High-confidence — 50 NM (modeled baseline). Routine detection of low-power tactical comms, AIS-class emitters, navigation radars at low altitude. RWR/ELINT receivers benefit from one-way path advantage over the emitter's two-way detection of itself. Source: RAND, FAS, RadarTutorial
EIRP-favorable — 150 NM (modeled baseline). High-EIRP emitters (search/track radars, ducted conditions, elevated targets). Outer envelope for the same passive receiver under favorable propagation. Source: CSIS, USNI Proceedings
Vendor tier model — three classes. The 50/150 NM ranges are conservative baselines defensible on the entry tier; mid- and high-end gear extends both envelopes meaningfully without changing the node footprint or power budget.
Entry tier — KrakenSDR ($1.2K, ~10W). 5-channel coherent SDR on a Pi-form-factor board, paired with a 5-antenna pentagon array on the mast head for direction-finding. Open-source-friendly, well-documented; sufficient for demonstration and many operational use cases at the modeled baseline ranges.
Mid tier — CRFS RFeye Node 100-18 ($25K+, ~50W). Ruggedized 1U chassis, 100 kHz – 18 GHz coverage. Professional integrator-grade. Achieves baseline ranges reliably plus better noise floor and TDOA networking for cross-fix.
High-end tier — meaningful range uplift on priority sites. All four options below remain within the Laulima HG-C1 node power budget (typically <1 kW combined with other modalities) and push HC to 60–120+ NM and EIRP to 150–250+ NM:
CRFS RFeye Array high-end (300-40 / 300-18) — 50–150W, 70–120+ NM HC / 180–300+ NM EIRP. Lowest power-per-capability in the high-end class. Best fit for broad deployment across the 251 sites.
Rohde & Schwarz DDF550 wideband DF + COMINT/CESM suite — 200–450W (core), 60–110+ NM HC / 150–250+ NM EIRP. Outstanding multipath immunity in maritime environments; high POI on agile/short signals.
L3Harris Rio Scalable SIGINT — 180–650W depending on config, 60–100+ NM HC / 150–240+ NM EIRP. Multi-channel coherent with beamforming; tunable channel count for SWaP optimization.
Elbit NATACS 2020 Naval Tactical COMINT/DF — 300–700W, 65–105+ NM HC / 160–260+ NM EIRP. Battle-proven for dense/agile maritime spectra; ideal where modern comms emitters dominate.
Alaris DF array front-end (e.g., DF-A0088) — +20–80W, pairs with any of the above. Rugged maritime 5-element interferometer optimizing real-world DF quality and near-horizon performance. Recommended mast-top front-end for premium-tier nodes.
Deployment philosophy — mixed fleet. Network-wide entry-tier KrakenSDR provides credible baseline coverage and demonstrates the architecture. Mid-tier CRFS Node 100-18 on the ~30 highest-value sites (Bokak, Wake, Tobi, Helen Reef, capital nodes, Sapwuahfik triangulation anchors). High-end CRFS Array, R&S DDF550, or Elbit NATACS 2020 on the 5–10 priority strategic anchors where extended HC range and superior agile-signal handling materially expand the operational envelope. The atlas's 50/150 NM modeled ranges represent the broadly-deployable baseline, not the upper bound. Source: vendor datasheets; conservative ducting/propagation modeling
Frequency-band extensions visualized on the map. The atlas now exposes three additional SIGINT layers beyond the VHF/UHF baseline, each with its own color-keyed coverage circle and (when triangulation is on) multi-source overlap contours:
SIGINT High-end VHF/UHF — 100 NM. The uplift envelope when CRFS Array high-end / R&S DDF550 / L3Harris Rio is deployed instead of entry-tier. 2× the baseline range, same physics, just better noise floor and processing.
HF SIGINT — 250 NM. 3–30 MHz, propagating via skywave/groundwave well beyond VHF horizon. Reach is 4–5× the baseline. Captures distant vessel comms and HF maritime transmissions that VHF receivers can't see. Reference hardware: R&S HF DF extensions or dedicated HF channels added to a DDF550 chassis (+50–200W).
SATCOM / L-band — 150 NM. ~1–2 GHz monitoring of vessel SATCOM uplinks. Captures "dark" vessels that have gone quiet on VHF/AIS but still use Inmarsat / Iridium / VSAT. CRFS or R&S higher-band heads, or dedicated L-band DF arrays (+100–400W).
Software/processing layers (not separately visualized — they overlay any active receiver):
Wideband ELINT (radar emitter focus) — precise PRI / pulse-width / scan-pattern measurement. Major boost for vessel classification. +50–300W shared with base front-end.
RF Fingerprinting / Specific Emitter Identification (SEI) — AI/ML layer that identifies individual emitters by hardware signatures, dramatically improves tracking persistence and anomaly detection. +20–100W compute overhead. Pure software addition over time.
Passive radar / bistatic exploitation — coherent processing on existing receivers detects vessels via reflections of broadcast/illuminator signals. Detects truly emission-dark targets within tens of NM. +50–200W mostly processing.
Drone-deployed SIGINT (future F10 payload) — miniaturized SDR/ELINT head on the F10 dynamically extends collection at node-cued contacts. Node provides bearing/classification, F10 closes for high-confidence ID at <10 NM.
Power reality. Even a fully-loaded premium node (high-end VHF/UHF + HF + SATCOM + ELINT + SEI + Alaris front-end + AIS/ADS-B/EO-IR + drone dock) draws <1.5 kW continuous. The Laulima HG-C1's 121 kWp solar + 264 kWh battery comfortably supports this with large reserves — power is not a constraint, even for the most fully-instrumented sites.
Local
EO/IR + acoustic direct — 10 NM. Mast-mounted gyrostabilized PTZ — reference hardware is FLIR M364C XR maritime thermal/visible (~9 kg, $30K-class, 640×512 cooled thermal) for primary, or Axis Q6135-LE visible-only PTZ ($3K-class) as a daylight-only alternative. Provides 24/7 perimeter watch while the F10 is in dock. The airborne F10 EO/IR extends this envelope when launched. Plus direct-path hydroacoustic where seafloor-cabled hydrophone arrays are practical. Surface horizon-limited; subject to weather, sea state, and biologics. Source: NVL/NVESD Johnson criteria; FAS/USNA acoustic propagation
Hydroacoustic Convergence Zones
30 / 60 / 90 NM rings. SOFAR channel sound refraction returns acoustic energy to the surface in narrow annular rings, not a continuous disk. Three-CZ detection (~90 NM) is a documented design target for tactical towed arrays. Available only at deep-water-adjacent sites (126 of 246 nodes — atolls with steep external bathymetry, volcanic islands without continental shelf). Lagoon-bound and shelf-locked nodes get only direct-path acoustic. Source: FAS, USNA reference; per the document, ~20-30 nm intervals (closer in cold water/Arctic)
Drone area of coverage (Skydio F10-class)
One flight budget — 120 NM total per sortie. The model treats the F10 as having a single per-sortie distance capability of 120 NM, executed in whichever direction the operator commands. Whether the drone goes out and returns or transits one-way to a recharge node, it flies the same total distance; only the geometry of the mission changes.
Surveillance / patrol mission (out-and-back from one node): 60 NM is the farthest outward boundary the drone can reach and return from with reserve. The full 120 NM budget = 60 NM out + 60 NM back. Closer-in patrols use a slice of the budget for the transit and spend the remainder on circuitous surveillance patterns, low-altitude inspection, or extended on-station time.
Hop transit mission (one-way to a recharge node): 120 NM is the maximum point-to-point distance the drone can fly between charging nodes with reserve. This is the absolute max edge length between two connected nodes.
Math anchor — F10 published envelope. 90-min flight time × 100 mph cruise (Skydio's stated top speed) = 150 statute miles theoretical max distance, ≈ 130 NM. Subtract a 12% reserve (~11 min for landing approach and contingencies, ample for an autonomous dock recovery): useful budget ≈ 132 mi ≈ 115 NM, rounded to 120 NM for clean modeling. Round-trip patrol = 60 NM each way. One-way hop = 120 NM. Source: Skydio F10 announced specs (Ascend 2025); 90-min minimum endurance, 100 mph published top speed
Why transit shows as 2-D ellipses, not lines: a drone transiting from node A to node B with hop budget 120 NM and direct A-B distance 40 NM has 80 NM of fuel slack to spend deviating from the straight path. It can visit any point P satisfying |AP| + |PB| ≤ 120 NM — geometrically, an ellipse with foci A and B. The shorter the direct distance between two nodes, the wider the ellipse (more deviation budget); near the 120 NM limit, the ellipse narrows to a thin lens. The patrol case is the same geometry where A = B — a circle of radius 60 NM (since 2|AP| ≤ 120 ⟹ |AP| ≤ 60). One unified geometry, two operational uses. Source: ellipse reachability under linear flight-time budget
Platform notes: Skydio F10 fixed-wing, autonomous launch/landing from a Skydio Dock, currently in early access (H1 2026). The "90+ min" published endurance suggests production may exceed 90 min, in which case the budget could grow further; we anchor on the published minimum to stay defensible. The F10 carries its own EO/IR sensor package, so when the drone is airborne it serves as both the cued investigation asset and the long-range extension of the stationary EO/IR coverage modeled separately.
The CONOPS implication: isolated far-edge nodes (Wake, Bokak, Helen Reef, Tobi, Kapingamarangi) have surveillance reach of one 60 NM out-and-back circle — that is the per-node limit when no recharge is reachable. Clustered island chains (the Yap-C corridor, the Mortlocks, the Marshall Ralik chain, the southern Caroline outers) form long hop-connected components where fleet coverage extends as a continuous 2-D band — far beyond what node counts or single-circle envelopes suggest. The network's surveillance reach is graph-theoretic; connectivity geometry matters more than counts.
Sustainment loop: drones charge from each node's microgrid (~3-5 kWh per sortie); AWG-supplied freshwater rinse on recovery for salt mitigation. The same plumbing that produces community drinking water also extends drone fleet availability — the corrosion problem that kills most marine drone programs is solved by dual-use AWG.
Coverage outline + area calculation: the atlas computes a binary occupancy grid over the network bounding box (each cell tested against all patrol circles and hop ellipses), totals the covered cells to give square miles, and runs a marching-squares contour extraction to draw the outer perimeter as a bold outline. Grid resolution adapts from 1.5 NM (single cluster) to 6 NM (network-wide) — the area number is accurate to within ~3%. With nothing selected, you see the entire mosaic's coverage; clicking a node restricts to that node's hop-connected component.
Greedy max-coverage optimization
Network ranking solves the actual maximum coverage problem. Given a budget of N nodes from the 251-site candidate pool, the algorithm returns the N nodes that maximize total drone area of coverage (union of patrol envelopes + hop transit ellipses at 120 NM hop / 60 NM patrol).
Algorithm: classical greedy. At each step pick the candidate node whose addition adds the most new covered cells to a binary occupancy grid (3 NM resolution, ~360,000 cells across the network bounding box). Maximum coverage is NP-hard in general but submodular and monotone, so greedy achieves a (1 − 1/e) ≈ 63% optimality guarantee — the best polynomial-time bound known. Result is deterministic given the parameters.
What you'll see: the top-N slider controls the budget. The first ~30 picks each add ~10,000 sq mi (isolated far-edge nodes contributing fresh patrol circles from empty); marginal gain drops sharply after that — rank 80 adds ~1,500 sq mi, rank 120 adds ~120 sq mi. The natural deployment knee is around 60-80 nodes for ~570K sq mi of coverage. Full 251-node deployment yields ~658K sq mi. Source: Nemhauser-Wolsey-Fisher (1978) submodular max approximation
Out of scope (explicitly not modeled)
Theater-class active radar (SPY-1/SPY-6, AWACS-class), forward-based X-band (TPY-2), Over-the-Horizon (JORN/Container) and ionospheric sky-wave systems are not deployable at the node footprint and bandwidth this network supports. The strategic regime (≥500 NM) is what the network contributes to via fusion and cueing of higher-tier theater assets, not what any individual node delivers.
The CONOPS argument: uniform deployable hardware, distributed siting strategy. Edge nodes carry the same passive sensor stack as backbone nodes — what differs is geographic position. Far-edge nodes extend the coverage blanket because their footprints don't overlap with anyone else's. The mosaic is what gives the network its theater reach.