Navigation in contested environments: operating beyond GNSS

Dependence on satellite navigation is one of the best-documented fragilities of modern infrastructure. GNSS signals reach the Earth's surface with power comparable to a candle seen from thousands of kilometers away: they are trivially blockable by physical obstacles, vulnerable to deliberate jamming, and susceptible to spoofing — the attack in which a hostile transmitter broadcasts plausible signals to lead a receiver to an invented position. The public electronic-warfare literature has treated the subject for decades; what has changed is the scale of exposure, as aircraft, ships, ground vehicles, and unmanned systems came to depend on the same fragile source.

Zendrix develops navigation for the world in which that source is not trustworthy. This article describes the principles of this research line — relative navigation, onboard sensor fusion, and verifiable external correction — at the level of method, with the usual commitment: declared capability is measured capability.

The principle: position is an estimate, not a fact

The first design error in navigation is treating position as a fact delivered by an oracle. Position is always an estimate with an error, and the correct engineering question is never "where am I?" but "where am I, with what uncertainty, and for how long has that uncertainty been growing?"

A system that can answer the second question can operate under GNSS denial in a degraded and honest way. A system that merely repeats the last position received from the satellite — a practice more common than admitted — is not navigating: it is memorizing. The difference appears in the first minute of jamming.

In the Zendrix research architecture, every platform maintains a continuous estimate of position and velocity built from its own onboard sensors, with uncertainty explicitly tracked. When GNSS is available and sound, it feeds the estimate as one source among others. When it stops being sound, the estimate continues — with growing error, declared and known. The system never feigns certainty it does not have: honest navigation under degradation is navigation that knows the size of its own ignorance at every instant.

Relative navigation: the world without a map

Absolute navigation — knowing exact geographic coordinates — is frequently treated as a requirement when it is not. For most autonomous operations, what matters is relative navigation: maintaining safe separation between neighboring platforms, following a trajectory defined relative to the starting point, and returning to a reference position. All of this can be done with useful accuracy without ever consulting a satellite.

Relative navigation combines measurements of own motion — acceleration, rotation, altitude from barometric and ground-distance sensors — with observations of the nearby environment. The error of this combination grows with time, and here the discipline of measurement matters: Zendrix research establishes measured drift floors in high-fidelity simulated environments, per maneuver class and per time horizon. A known, auditable maximum drift is an engineering property; an unknown drift is a delayed accident.

Verifiable external correction: landmarks that do not lie

The growth of relative error is countered with external corrections — but in a contested environment, every external source is also a suspect source. A spoofed navigation landmark is as dangerous as a spoofed satellite.

The Zendrix approach treats external correction as a problem of verifiable trust. Known, immutable reference points — visual landmarks in the terrain, cataloged fixed infrastructure — offer high-credibility corrections because their existence can be confirmed independently of the navigation channel itself. The fusion weights each correction by its declared and verified credibility: an observation of uncertain origin enters the estimate with low weight or not at all; an observation confirmed by multiple independent sources enters with high weight. And every accepted correction is recorded in the estimate's audit trail — the system knows, and declares, what it believed and why.

This design has a property we consider essential for defense infrastructure: degradation is gradual and measurable. Losing GNSS does not bring navigation down; one source is lost, the declared uncertainty grows within known bounds, and the available verifiable corrections hold the error. Losing the external corrections too does not bring the operation down either: it brings accuracy down to a declared floor, beyond which the system refuses the maneuver and retreats to safe behavior. Refusing on uncertainty limits is a capability, not a failure.

Dual use: from the contested field to the dense city

GNSS denial is not only a military scenario. Urban canyons take down satellite signals between tall buildings; mines and tunnels operate without coverage by definition; the interior of critical facilities blocks signal naturally; solar storms degrade the accuracy of an entire region for hours. Infrastructure inspection, logistics in dense yards, emergency response in urban areas, and agriculture under closed canopy face, in civilian form, the same physical problem that electronic warfare produces in hostile form.

That is why we treat resilient navigation as dual-use intelligence infrastructure: the same technology base that sustains defense operations in contested environments sustains civilian operations in obstructed ones. It is a research line in which sovereignty and economic utility coincide — and in which the alternative (depending exclusively on constellations we do not control) is strategically and commercially fragile.

The state of the research

Current results come from high-fidelity simulation with modeled sensors — noise, drift, and partial failures included — and deterministically reproducible environments. Drift floors are measured and frozen as acceptance criteria; negative controls verify that a system which merely memorizes the last known position does not cross the gate. The transition to physical validation will follow the method described in our publication on simulation as gate: the same software, the same rigor, the physical world at the end of the process.

The institutional position of Zendrix is direct. We do not promise perfect navigation where no system in the world offers it. We build and measure honest navigation for contested environments: estimates with declared uncertainty, corrections with verifiable credibility, degradation with known floors, and safe refusal when the limit arrives. This is how one prepares the national technology base to operate — with responsible autonomy and human authority intact — when the electronic sky stops being trustworthy.