Building an electronic order of battle from ELINT data

An electronic order of battle (EOB) is one of the most operationally consequential intelligence products a SIGINT organization produces. Where a conventional order of battle catalogs the ground, air, and maritime forces an adversary has deployed, the EOB catalogs their electromagnetic capabilities: which radars are operating, on which platforms, with what parameters, at what locations, and with what confidence. This article explains how ELINT collection data is transformed into a maintained EOB — from raw signal parameters through emitter characterization, specific emitter identification, platform association, and the database management workflows that keep the EOB current.

What an EOB is and why it matters

The electronic order of battle is a structured intelligence database whose fundamental unit is the emitter record. Each record describes a single electromagnetic emitter — in most operational contexts, a radar or electronic warfare system — and captures what is known about it: the technical parameters that characterize its waveform, the platform that carries it, its geographic operating area, its current operational status, and the confidence associated with each of these elements. The EOB is distinguished from a simple emitter observation log by its organizational structure: emitters are linked to platforms, platforms are linked to units, units are linked to formations, and the whole structure reflects the adversary's actual force deployment.

The EOB directly supports several operational functions. Electronic warfare planning depends on the EOB to identify which threat emitters must be suppressed, jammed, or avoided for a mission to succeed. Route planning uses EOB-derived radar coverage maps to identify flight or movement corridors that minimize radar exposure. Intelligence production uses the EOB to track the composition and disposition of adversary forces over time — the appearance of a fire-control radar at a new location is an operationally significant event. Electronic support measures (ESM) systems on aircraft and ships use the EOB as their primary threat library, matching observed emissions against known emitter types to generate threat warnings.

An EOB entry for a surface-to-air missile radar typically contains: the radar designation and type (acquisition radar, tracking radar, fire control radar, guidance uplink), the associated missile system and launcher, the parameter bounds for each measured characteristic (frequency range, PRI range, scan rate), the geographic operating area with positional uncertainty, the observed activity history (first detected, mode changes, periods of silence), and the confidence level for each element. Confidence is not uniform across entries: a fire-control radar observed fifty times by multiple collection assets has very different epistemic status than one observed once by a single ground station.

ELINT collection and parameter extraction

ELINT collection is conducted from ground stations, airborne platforms, and satellites, each with different geometries, coverage areas, and measurement capabilities. Ground-based collection stations provide continuous coverage of fixed areas and long dwell times that enable precise parameter measurement, but are limited by terrain masking and cannot be repositioned quickly. Airborne collection platforms — manned ISR aircraft and increasingly unmanned aerial vehicles — provide flexible coverage and favorable geometry for geolocation, but have limited on-station time. Satellite-based collection covers large areas continuously but has geometric limitations for geolocation and constrains the measurement of time-varying parameters.

The parameters extracted from each ELINT observation fall into two categories. Intentional parameters are those the radar designer chose: carrier frequency (or the frequency range if the radar is frequency-agile), pulse repetition interval (PRI), pulse width (PW), scan period, scan type (sector scan, continuous circular scan, conical scan, phased-array stare), and modulation characteristics such as linear frequency modulation (LFM) chirp or pulse-Doppler waveforms. These intentional parameters are what the emitter parameter library uses for type identification. Unintentional parameters arise from manufacturing tolerances and component aging: pulse rise time variations, carrier frequency offset and drift, amplitude ripple within the pulse envelope, and spurious sidebands. These unintentional parameters are the basis for specific emitter identification.

Measurement uncertainty is an inescapable feature of ELINT collection. A PRI measurement derived from a short collection opportunity, with poor signal-to-noise ratio and a single receiver, carries wide uncertainty bounds. The same parameter measured over a long dwell at high SNR from a calibrated receiver is far more precise. EOB databases that record only the parameter value without recording the measurement uncertainty produce misleading confidence in the data. Every parameter in an EOB entry should carry its measurement bounds, the collection geometry that produced them, and the number of observations on which they are based.

Emitter characterization

Emitter characterization is the process of assigning a type identity to an observed emitter — determining not just that a radar is present but what kind of radar it is. The primary method is parameter library matching: comparing the observed parameter set against entries in a database of known emitter types, each entry specifying the expected parameter ranges for that system. The match is evaluated as a similarity score that accounts for the measurement uncertainty of each observed parameter — an observed PRI that falls within the measurement uncertainty of the library value contributes a strong match signal even if it is not numerically identical.

Radar type identification distinguishes among the functional roles: early warning (long range, long PRI, wide beam, slow scan), acquisition/search (medium range, multiple PRIs, sector or circular scan), target tracking (narrow beam, fast update rate, range-gated PRIs), fire control (very precise, narrow beam, high update rate), and missile guidance (uplink characteristics, specific timing relationships with tracking radar). Each functional type has characteristic parameter combinations, and most library entries contain both the nominal values and the expected variation bounds for each parameter across the known production variants of each system.

EWIR — the Electronic Warfare Integrated Reprogramming data format used by Western armed forces — provides the standardized structure for emitter library entries used in airborne ESM systems. An EWIR entry captures the parameter bounds for each mode of operation (search mode, track mode, engagement mode), the threat warning conditions, the recommended electronic countermeasure responses, and the confidence information for each parameter. EOB databases for larger intelligence organizations use richer data models than EWIR entries, but the parameter structure is similar: mode-by-mode characterization with explicit uncertainty bounds and provenance.

Specific emitter identification and identity resolution

Specific emitter identification (SEI) extends type classification from "this is an SA-15 Tor radar" to "this is the specific SA-15 Tor radar with serial characteristics matching emitter track ID-4471." SEI is based on the unintentional modulation on pulse (UMOP) characteristics that arise from hardware manufacturing tolerances and component aging. Every transmitter produces slightly different pulse shapes, frequency centroids, rise time distributions, and amplitude profiles. These differences are small — typically fractions of a percent of nominal values — but stable and reproducible for a given hardware unit, making them a reliable fingerprint.

The SEI process begins with high-fidelity waveform collection at sufficient bandwidth to capture intra-pulse features. The key features extracted are: pulse rise time distribution (mean and standard deviation across many pulses), carrier frequency centroid and jitter spectrum, amplitude envelope shape and within-pulse ripple pattern, and phase noise characteristics. These features are compared against the SEI reference library — a collection of signatures from emitters that have been previously characterized, either from direct collection or from controlled measurements of captured hardware.

Identity resolution handles the case where a new observation is ambiguous between two or more candidate identities. The confidence score for each candidate is computed from the feature-space distance between the observation and the library entry, normalized by the within-class variance of that entry. A distance of 1.2 standard deviations produces a lower confidence than a distance of 0.3 standard deviations. When two candidates are nearly equally plausible, the system flags the ambiguity rather than forcing a single identification. The analyst reviews the ambiguity with access to the collection geometry, the historical track of each candidate, and any corroborating intelligence.

Platform association

Platform association is the process of linking individual emitter records to the specific vehicle, aircraft, or vessel that carries them. Most military platforms carry multiple emitters: a surface combatant typically carries a navigation radar, a surface search radar, a fire control radar, and one or more electronic warfare systems simultaneously. Associating these emitters with the same ship class — and ideally the same hull number — is essential for producing actionable EOB entries.

The primary method of platform association is co-emission analysis. If emitters A, B, and C consistently appear together in time and geographic position, they are likely collocated on the same platform. Co-emission is not merely about simultaneous operation — emitters on a complex platform may have different operational duty cycles. The association signal comes from consistent co-location over multiple collection opportunities: a fire control radar and a navigation radar that are always within 50 meters of each other across twenty separate collection events are almost certainly on the same vessel.

Movement correlation provides a secondary association method. If an emitter's geographic track over time follows a trajectory consistent with a specific platform type — a ship following a plausible sea lane at plausible ship speeds — the movement pattern constrains the platform association. When the movement track can be correlated with a specific vessel track from another sensor (maritime patrol aircraft radar, ADSB equivalent for naval vessels), the association becomes multi-source confirmed.

COMINT-ELINT fusion provides a third association path. Communications emitters on the same platform as radar emitters often share geographic position and movement track. If COMINT collection has associated a specific radio net with a particular vessel or aircraft, and an ELINT emitter track precisely follows the same movement, the association can be inferred. Multi-source confirmation — agreement between co-emission, movement correlation, and COMINT association — produces the highest-confidence platform associations in the EOB.

EOB database schema and update workflows

The EOB database schema must capture the full epistemic structure of the intelligence: not just the current best estimates of each parameter but the full history of observations that produced those estimates, the confidence trajectory over time, and the provenance chain linking each EOB element to the specific collection events that support it. A minimal schema has three primary tables: the emitter record (emitter ID, type identification, parameter bounds, last observed, operational status, confidence), the platform record (platform ID, platform type, class, hull or tail number, associated unit), and the association table (emitter-platform links with association confidence, evidence type, and evidence timestamps).

EOB update workflows define how new collection events enter the database. An automated update pipeline ingests new ELINT observations, runs the parameter library match and SEI processes, proposes type identifications and emitter track correlations, and queues the results for analyst review above a confidence threshold. Below the threshold, the observation is flagged as a new candidate entry. Analyst-reviewed updates carry an authority record — which analyst approved the update, at what time, and on what basis — supporting audit and reprocessing if the library is later revised.

Version control is essential for operational EOB databases. At any given time, different downstream consumers may be working from different versions of the EOB: an aircraft that departed on a long mission carries the EOB version that was loaded at departure, while the ground station has a more current version. Version management must track which consumers have which version, provide a delta format for efficient updates, and support rollback if an update later proves incorrect.

Confidence scoring and attribution uncertainty

Confidence scoring in an EOB database operates at multiple levels: parameter-level confidence (how precisely is each technical parameter known?), identity-level confidence (how certain is the type identification?), platform association confidence (how certain is the emitter-to-platform link?), and entry-level confidence (a composite score that drives dissemination decisions). Each level is computed and stored independently because different consumers need different confidence cuts: a route planner needs high-confidence threat emitter positions; an analyst building a long-term order of battle assessment may use lower-confidence entries as evidence for trends.

A Bayesian update model is the appropriate framework for EOB confidence management. Each new observation of an emitter updates the probability distribution over its parameter values, with the update weighted by the observation quality (SNR, collection duration, geometry). Confidence increases with the number and quality of observations, and decays with time as the operational situation may have changed. A fire-control radar last observed six months ago has different epistemic status than one observed last hour: the historical observation still constrains the emitter's type and operating area, but its current status (active, repositioned, destroyed) is unknown.

Conflicting report resolution arises when two collection sources produce significantly different parameter values for the same emitter. The resolution process begins with checking whether the discrepancy can be explained by collection geometry — an aspect-angle difference can produce large apparent differences in measured scan rate or PRI for some radar types. If geometry does not explain the discrepancy, the sources' measurement methodologies are compared and reliability weights assigned. The resolved parameter bounds are wider than either source alone, explicitly acknowledging the unresolved uncertainty. The conflict record is retained so future analysis can revisit the resolution if additional observations clarify the situation.

Dissemination with uncertainty metadata is the final step. EOB extracts sent to downstream consumers — ESM systems, mission planning tools, intelligence reports — should carry the confidence scores and uncertainty bounds for each field, not merely the point estimates. A mission planning tool that receives a radar position as a single grid reference treats that position as known; a tool that receives the position with a 5-kilometer error ellipse can compute the appropriate standoff distance. The difference between these two behaviors is operationally significant when the threat system has a lethal range measured in tens of kilometers.