EW Math Guide Module

Direction Finding and Emitter Geolocation

Angle-of-arrival mathematics for EW receivers: amplitude comparison, Watson-Watt, pseudo-Doppler, phase interferometry, array AOA, triangulation and error geometry.

DF Measures Direction Before It Measures Position

Direction finding estimates the bearing or angle of arrival of an intercepted emitter. A single DF sensor normally gives a line of bearing, not a position. Position appears when several bearings, time differences, frequency differences or motion-assisted measurements are combined.

In EW this matters because the receiver is often passive. It does not illuminate the target like radar. It listens to hostile or unknown emissions, extracts signal parameters, estimates the wavefront direction and sends that bearing to an operator, tracker or geolocation processor.

Learning target: read DF formulas as engineering measurements, calculate bearing error, understand ambiguity, and connect AOA measurements to emitter location uncertainty.

A DF sensor converts antenna response, phase or Doppler modulation into a bearing line. Several bearing lines intersect to form an estimated emitter location and an error ellipse.

AOA Geometry and Bearing Lines

An angle-of-arrival measurement is a geometric constraint. From a sensor position, the emitter must lie somewhere along the measured bearing, with an angular uncertainty caused by noise, multipath, calibration error and finite aperture.

Bearing line model

θ = atan2(y - ys, x - xs) + nθ
u(θ) = [cosθ, sinθ]T
p = ps + r u(θ), r > 0

Formula reading: the emitter position is p = [x, y]. The DF sensor is at ps = [xs, ys]. The measured bearing θ is the true geometric angle plus angular error nθ. The unit vector u(θ) points from the sensor along the measured line of bearing.

Worked Example: bearing line from one sensor

Sensorps = (0 km, 0 km)
Measured bearingθ = 35°
Line of bearingp(r) = r[cos35°, sin35°] = r[0.819, 0.574]
MeaningThe emitter may be anywhere on that line until another constraint is added.

Amplitude-Comparison DF

Amplitude DF compares received power in two or more directional antenna patterns. It is simple and fast, but accuracy depends strongly on antenna pattern calibration, polarization, multipath and signal fading.

Ratio and monopulse-like error signal

P1(θ) = PrG1(θ)
P2(θ) = PrG2(θ)
RdB = P1,dB - P2,dB
ε = PR - PLPR + PL

Formula reading: each channel sees the same incident signal power Pr, multiplied by its directional gain pattern. The dB ratio removes much of the unknown emitter power. The normalized difference ε is a left-right angle error signal: zero means the emitter lies on the electrical centerline, positive or negative tells which side.

Worked Example: calibrated two-beam ratio

Measured channel powersPR = -47 dBm, PL = -53 dBm
RatioRdB = -47 - (-53) = +6 dB
Calibration table+6 dB corresponds to +8° off boresight for this antenna pair
Operational noteThe number is valid only for the calibrated frequency, polarization and installation.

Watson-Watt DF

Watson-Watt systems use two orthogonal directional channels, often represented as sine and cosine patterns, plus an optional sense channel to remove the 180° ambiguity. The bearing comes from the arctangent of the channel voltages.

Orthogonal channel angle

Vx = K E cosθ
Vy = K E sinθ
&hat;θ = atan2(Vy, Vx)

Formula reading: Vx and Vy are the two orthogonal DF channel voltages. K is receiver gain and E is field strength. Because both channels share K and E, the angle is found from their ratio. atan2 is used instead of a simple tangent so the correct quadrant is retained.

Worked Example: Watson-Watt bearing

Measured voltagesVx = 0.62 V, Vy = 0.78 V
Bearing&hat;θ = atan2(0.78, 0.62) = 51.5°
Sense antennaConfirms front hemisphere instead of the reciprocal 231.5° solution.

Pseudo-Doppler DF

Pseudo-Doppler DF electronically switches around a circular antenna array. The phase of the resulting frequency modulation indicates the incoming wave direction. It is common in compact VHF/UHF DF receivers because it can emulate a rotating antenna without mechanical rotation.

Doppler modulation from circular motion

fd,max = 2πfrotaλ
fi(t) = f0 + fd,maxcos(2πfrott - θ)

Formula reading: a is the effective rotation radius, frot is switching/rotation rate, and λ is wavelength. The larger the circular aperture relative to wavelength, the stronger the Doppler modulation. The unknown arrival angle θ appears as the phase of that modulation.

Worked Example: pseudo-Doppler deviation

Inputsf = 150 MHz, λ = 2.0 m, a = 0.25 m, frot = 500 Hz
Maximum deviationfd,max = 2π × 500 × 0.252.0 = 393 Hz
MeaningThe receiver estimates the phase of a 500 Hz modulation with about 393 Hz peak deviation.

Phase Interferometer DF

A phase interferometer measures the phase difference between spatially separated antennas. It can be very accurate because phase can be measured finely, but baselines longer than half a wavelength introduce ambiguous angle solutions unless multiple baselines or coarse sensors are used.

Two-element phase baseline

Δφ = 2πd sinθλ
&hat;θ = sin-1λΔφ2πd
d ≤ λ2 for unambiguous ±90°

Formula reading: d is antenna spacing. A wave arriving from angle θ reaches one antenna before the other, creating path difference d sinθ. Dividing that path by wavelength gives phase. Since measured phase wraps every 2π, long baselines create several possible angles.

Worked Example: phase to bearing

Inputsf = 1 GHz, λ = 0.30 m, d = 0.15 m, Δφ = 90° = π2 rad
Angleθ = sin-10.30 × 1.5712π × 0.15 = 30°
AmbiguityBecause d = λ2, this baseline is just unambiguous over the forward hemisphere.

Array-Based AOA and Beamformer Scanning

Digital EW receivers can estimate direction by comparing the received vector across many antenna elements. The same steering-vector mathematics supports conventional beam scanning, maximum likelihood estimators and subspace methods such as MUSIC.

Steering vector and beam scan

a(θ) = [1, ejkd sinθ, ..., ej(M-1)kd sinθ]T
PBF(θ) = |aH(&theta)x|2
k = λ

Formula reading: a(θ) predicts the phase progression across an M-element array for a plane wave from angle θ. x is the measured complex channel vector. The beamformer tests candidate angles and picks the angle whose steering vector best matches the received data.

Worked Example: array aperture and beamwidth

ArrayM = 8 elements, spacing d = λ2
ApertureD ≈ (M - 1)d = 3.5λ
Approximate beamwidthθBW0.886λD = 0.253 rad = 14.5°
MeaningLarger aperture narrows the search beam and improves angular discrimination.

AOA Geolocation by Triangulation

Two or more DF sensors can geolocate an emitter by intersecting bearing lines. In practice the lines do not meet exactly, so least squares or maximum likelihood finds the position that best fits all bearing measurements.

Least-squares bearing residual

ri(p) = wrap(θi - atan2(y-yi, x-xi))
p̂ = arg minp Σi ri2(p)σθ,i2

Formula reading: each sensor contributes an angular residual: measured bearing minus predicted bearing from the trial emitter position. The residual is wrapped so 359° and -1° are treated correctly. Measurements with smaller angular variance receive more weight.

Worked Example: two-bearing intersection

Sensor A(0 km, 0 km), bearing 35°
Sensor B(20 km, 0 km), bearing 145°
IntersectionSolving the two line equations gives approximately (14.3 km, 10.0 km)
Geometry warningIf the two bearings are nearly parallel, the position error grows rapidly.

Error Ellipse, CEP and DF Accuracy

A one-degree DF error can be small nearby and huge far away. The lateral position uncertainty grows approximately linearly with range. Geolocation geometry is therefore as important as receiver quality.

Cross-range error and covariance

σcross ≈ R σθ
CEP50 ≈ 0.59(σmajor + σminor)
GDOP: weak crossing angle → large error ellipse

Formula reading: angular error must be in radians. A bearing uncertainty creates a wedge. At range R, the cross-range width is roughly R times the angular standard deviation. The error ellipse summarizes how all sensor geometry and measurement errors map into position uncertainty.

Worked Example: one-degree bearing uncertainty

RangeR = 50 km
Angle standard deviationσθ = 1° = 0.01745 rad
Cross-range errorσcross ≈ 50 × 0.01745 = 0.87 km
LessonGood DF accuracy still needs good geometry and enough baseline separation.

Where TDOA and FDOA Fit

DF/AOA is only one passive geolocation family. Time difference of arrival creates hyperbolas; frequency difference of arrival creates Doppler contours. Modern EW geolocation often fuses AOA, TDOA, FDOA, signal ID and platform motion.

TDOA

Δtij = ||p-pi|| - ||p-pj||c0

Formula reading: a time-difference measurement says the emitter is closer to one sensor than another by c0Δt. The locus of constant range difference is a hyperbola in 2-D.

FDOA

Δfij = -f0c0[(vi·ui) - (vj·uj)]

Formula reading: FDOA uses relative Doppler shift. Moving receivers see slightly different frequencies depending on their velocity projection toward or away from the emitter.