The Dependency That Was Always There and Is Now Urgent
GPS entered operational service with the US military in the 1990s and transformed precision warfare so thoroughly that requirements writers have treated it as infrastructure — background like AC power — rather than a system that could be denied. Precision-guided munitions, autonomous vehicle navigation, network time synchronization for data correlation, and targeting geometry calculations all depend on GPS signals. The dependency deepened as AI-enabled systems multiplied the uses: fusion algorithms that correlate multi-INT feeds rely on synchronized timestamps accurate to the microsecond, autonomous ground vehicles require continuous sub-meter position updates, and counter-UAS fire control depends on relative navigation for intercept geometry.
The adversary response has been well documented in unclassified reporting and government testimony. Russia demonstrated GPS jamming in the Baltic region during NATO exercises and in areas adjacent to operations in Ukraine. The Defense Intelligence Agency and NATO allies have described these capabilities in public summaries. China has developed parallel GPS jamming and spoofing capabilities assessed as operationally relevant in the Pacific. The tactical consequence is that GPS degradation at the formation level is a planning assumption, not a worst-case scenario, for high-end conflict against peer adversaries.
What has changed in the last several years is not the threat, which has been understood since at least 2012, but the AI-enabled capability stack that is now at risk. Legacy GPS-guided weapons could tolerate interruption through inertial backup — a terminal phase was brief enough that moderate drift was acceptable. An autonomous system making continuous maneuvering decisions about obstacle avoidance, targeting correlation, and route selection cannot tolerate extended navigation uncertainty. The quality-of-service requirement that AI-enabled autonomy places on navigation is qualitatively different from what precision-guided weapons required, and it creates a harder GPS dependency just as the threat to GPS has matured.
What A-PNT Actually Means
The DoD's A-PNT portfolio is an umbrella of complementary approaches managed across multiple program offices. The architecture has three conceptual layers.
The first is hardened GPS reception. GPS III satellites include stronger signals and improved anti-jam characteristics compared to GPS IIR/IIF. The GPS M-Code signal provides higher power and encrypted access that consumer and standard military receivers cannot use — it is designed specifically for environments where adversaries are attempting to jam or spoof civilian GPS bands. Military-grade M-Code receivers have been a persistent acquisition bottleneck because of cost and constrained production volumes, which limits the density of GPS-hardened capability in operational formations.
The second layer is alternative positioning sources that function independently of GPS. Inertial Navigation Systems with MEMS-based IMUs have improved significantly in size, weight, drift rate, and cost over the past decade. Terrain-referenced navigation, which compares sensor-derived terrain data to pre-stored maps, provides position updates without any radio frequency signal and is difficult to jam or spoof. Signals of opportunity — using LEO commercial satellite signals, eLoran where infrastructure supports it, and visual odometry for ground vehicles in mapped environments — provide additional independent position and timing sources. No single alternative matches GPS for accuracy and continuity, but together they can bound position uncertainty to operationally acceptable levels.
The third layer is fusion: algorithms that weight, combine, and error-correct inputs from multiple sources in real time. A system that switches from GPS to INS when GPS is denied gets the worst of both — it loses accuracy on transition and cannot gracefully degrade through partial signal degradation. A properly architected multi-source navigation fusion engine continuously weights all available signals, detects spoofing through cross-source consistency checking, and maintains bounded position uncertainty rather than either false confidence or total loss. This is fundamentally an AI engineering problem: it requires probabilistic state estimation, anomaly detection, and rapid adaptation to changing signal availability.
The Spoofing Problem Is Worse Than the Jamming Problem
Brute-force GPS jamming is detectable. Receivers can identify when signal levels have dropped below normal, report an unreliable fix, and trigger fallback to INS. Formation commanders know they are operating degraded.
Sophisticated spoofing is more dangerous because it is silent. A spoofing signal feeds plausible false GPS coordinates to receivers, which report a confident fix — at the wrong location. The system believes it knows where it is. For an AI-enabled targeting system, a spoofed position fix can corrupt an entire kill chain calculation without triggering any alert. For an autonomous vehicle, it can direct the platform into restricted areas or away from its objective without the operator's awareness that navigation has been compromised.
Detecting spoofing requires exactly the architecture that honest multi-source navigation provides: comparing the GPS-derived position against INS-derived position, against terrain reference, against any other available source. When the cross-source comparison fails consistency checks, the system can flag the discrepancy before acting on potentially false data. This is not exotic technology — it is a straightforward application of sensor fusion with consistency monitoring — but it requires that the navigation architecture be designed to accept and continuously compare multiple sources from the beginning, not patched in later.
Army A-PNT Program Architecture
The Army's A-PNT acquisition programs are managed by PM PNT (Project Manager Positioning, Navigation, and Timing), which sits under Capability Program Executive Intelligence and Spectrum Warfare (CPE ISW), formerly PEO IEW&S. The A-PNT Cross-Functional Team under Army Futures Command developed requirements across the portfolio during its active years; it completed its charter and was repurposed to the All-Domain Sensing Cross-Functional Team in early 2024, reflecting that A-PNT had matured from requirements development into an active acquisition program with PM PNT as the execution authority. The breadth of the A-PNT problem — covering ground vehicles, aircraft, UAS, soldier systems, and the tactical network that ties them together — is why the organizational structure emphasized cross-platform coordination from the start.
Active acquisition threads managed by PM PNT include the Mounted Assured PNT System (MAPS), which provides a vehicle-level integration of multiple navigation sources including GPS, INS, and alternate signals into a standardized interface for vehicle platforms. MAPS Generation II received full-rate production approval in early 2025 and continues to expand across the vehicle fleet. The Dismounted Assured PNT System (DAPS) addresses the soldier-level requirement. The Army's Integrated Tactical Network incorporates A-PNT requirements for network time synchronization to ensure that data correlation across the tactical network does not depend solely on GPS timing.
The organizational co-location of TITAN (AI-enabled ISR fusion) and PM PNT (A-PNT hardware) under CPE ISW is architecturally logical — an AI-enabled targeting system that fuses intelligence feeds at corps echelon has a direct dependency on reliable position and timing inputs. Whether A-PNT-hardened navigation has been specified as a formal TITAN requirement in solicitation documents, or is addressed through the MAPS interface standard at the platform level, the dependency exists: a fusion engine whose data streams are timestamped to a GPS clock it cannot verify is vulnerable to exactly the spoofing attacks the A-PNT architecture is designed to detect.
DARPA's Contribution: Adaptable Navigation Systems
DARPA's Adaptable Navigation Systems (ANS) program addresses the navigation gap with two concurrent efforts. PINS (Precision Inertial Navigation Systems) develops inertial measurement units that are fundamentally more accurate and less prone to drift than current MEMS designs — if the dead-reckoning baseline is better, every multi-source system that incorporates it performs better. ASPN (All Source Positioning and Navigation) addresses the fusion side: algorithms and plug-and-play integration architectures that let a navigation processor accept inputs from any available sensor without requiring custom integration for each new sensor type. DARPA's stated goal for ANS is "GPS-quality PNT to military users regardless of the operational environment" — an acknowledgment that the problem is not to find one good GPS substitute but to build an architecture that can draw on whatever signals exist in the local environment.
ANS represents the research boundary because it is deliberately working on the worst cases: environments where GPS is fully denied, where known alternative signals are also actively contested, and where the navigation system must use passive environmental features — terrain profiles, magnetic anomaly maps, gravity gradients — to maintain position without emitting any signal at all. The operational relevance is that the Pacific and European contested environments include scenarios at that level of denial.
The Timing Dependency That Is Often Overlooked
The focus on positioning tends to overshadow the timing dependency, which is equally critical for AI-enabled systems.
GPS provides a timing reference of extraordinary precision — receiver clocks synchronized to GPS time are accurate to tens of nanoseconds. Tactical networks use this timing reference for data correlation across sensors that may be kilometers apart and communicating through multiple relays. An AI fusion engine correlating SIGINT, IMINT, and HUMINT-derived data to build a targeting picture requires that all data streams be timestamped to a common reference. If the timing reference is GPS and GPS is denied, data correlation degrades or fails even if each sensor is still collecting.
Network timing resilience is an under-resourced component of the A-PNT problem. Solutions include Chip-Scale Atomic Clocks (CSACs) that maintain GPS-derived timing through an outage using local oscillators, two-way timing through alternative networks, and timing distribution architectures that can transition between multiple references without losing correlation accuracy. Contractors building AI-enabled systems for operational use need to specify their timing architecture as explicitly as their navigation architecture — and the acceptable drift rate during a GPS outage must be specified relative to the mission's data correlation requirements, not as a generic number.
What Program Offices Are Signaling to Industry
Several acquisition signals point to how A-PNT requirements are being operationalized.
The Robotic Combat Vehicle program, with its autonomous navigation requirement, operationally depends on A-PNT resilience — an unmanned ground vehicle operating in denied environments cannot use GPS as its sole navigation source. The Air Force Collaborative Combat Aircraft, designed for contested airspace, has navigation resilience as an inherent operational constraint on the system's usefulness. The Marine Corps' Ground Based Air Defense programs require accurate position for both the fire-control system and the threat track; counter-UAS intercept geometry breaks down when position is uncertain.
Congressional attention to GPS resilience has grown as the threat has become better documented. The GPS Resiliency Report Act (S.2277, 119th Congress, introduced 2025) would direct DoD reporting on GPS vulnerabilities and alternatives and the pace of A-PNT fielding. Whether or not specific NDAA provisions have enacted reporting requirements, the pattern across multiple successive authorization cycles reflects a consistent Congressional concern that the A-PNT capability developed under DARPA and Army programs has moved into acquisition faster than it has moved into formations.
The Contractor Checklist
Companies competing in autonomous systems, AI-enabled targeting, precision fires, and tactical networking need to treat A-PNT compliance as a procurement requirement rather than a capability differentiator. Program offices are writing it into solicitations; treating it as an enhancement to add at LRIP is a losing strategy.
The technical requirements that matter are:
- Multi-source navigation input architecture: INS, GPS, terrain reference, signals of opportunity — not sequential fallback but continuous parallel input
- Probabilistic sensor fusion: continuous weighting based on real-time signal quality, not binary GPS/INS switching
- Spoofing detection: cross-source consistency checking with configurable alarm thresholds
- Graceful degradation specification: explicit position uncertainty bounds for each degradation scenario (GPS degraded, GPS denied, INS-only, terrain-reference-only)
- Timing resilience: defined holdover accuracy during GPS timing outage, with explicit correlation performance implications
- Position data interface: compliant with the Army's MAPS architecture and the emerging joint A-PNT interface standards so that position data can be shared across formation elements
For AI-enabled systems specifically, the integration requirement extends into the inference layer. A targeting algorithm or autonomous navigation decision loop that was architected assuming GPS availability needs to be re-examined for how it handles position uncertainty increases. An algorithm that assumes sub-meter position accuracy and breaks down at ten-meter uncertainty was not designed for the contested environment — it was designed for a test range.
Sources and further reading
- PM PNT (Project Manager Positioning, Navigation, and Timing) program pages under CPE ISW (Capability Program Executive Intelligence and Spectrum Warfare): cpeisw.army.mil/pm-pnt/; MAPS and DAPS acquisition documentation on sam.gov
- DARPA Adaptable Navigation Systems (ANS) program: darpa.mil/research/programs/adaptable-navigation-systems
- Joint Publication 3-14, "Space Operations" (current edition) — addresses position, navigation, and timing at the joint force level and identifies PNT as a mission-essential function across joint operations
- DoD GPS Program Executive Office, GPS III and GPS IIIF program documentation, available through Space Systems Command
- OUSD(R&E) Trusted and Assured Microelectronics (T&AM) program documentation — addresses hardware security in PNT receivers
- GPS Resiliency Report Act (S.2277, 119th Congress, 2025) — introduced legislation addressing DoD GPS vulnerability reporting and alternative PNT fielding
- DHS Resilient PNT Reference Architecture, and DoT National PNT Advisory Board documentation — civilian side of the A-PNT policy and standards ecosystem
Spartan X's edge compute work — specifically BRIC's architecture for AI inference in DDIL environments — intersects directly with the A-PNT challenge. Deploying AI-enabled capability in contested theaters requires that the full system stack, navigation and timing included, functions reliably when adversaries target the signals the system depends on most. The experience of integrating inference pipelines against intermittent, degraded, and spoofed data inputs is the same engineering discipline whether the data is communications, sensor feeds, or navigation.
