VHF/UHF direction finding for tactical SIGINT: algorithms, networks, and mobile deployment

Direction finding (DF) in the VHF and UHF bands remains the most operationally reliable method for locating tactical emitters -- push-to-talk radios, data links, drone control channels, and short-range command nets. Unlike TDOA systems that require nanosecond-precision time synchronization across widely separated receivers, a single DF platform can contribute a useful bearing line with nothing more than a calibrated antenna array, a coherent multichannel receiver, and a few milliseconds of signal intercept. Add a second platform and you have a fix. Add a third and you have a redundant, geometry-robust geolocation. This article examines the physics and engineering behind tactical VHF/UHF DF: from antenna array selection and DF algorithm mathematics to DF network architecture for coordinated operations, multipath mitigation in complex terrain, and integration with the SIGINT collection chain.

VHF/UHF DF in tactical SIGINT: frequency bands and operational contexts

The VHF band (30--300 MHz) and UHF band (300 MHz--3 GHz) together cover the vast majority of tactical radio traffic encountered in land warfare. VHF is the traditional band for military voice nets, manpack radios, and vehicle-mounted command sets, with propagation characteristics -- including ground wave at the low end and near-line-of-sight at the high end -- that support communication ranges of 5--50 km depending on terrain and antenna height. UHF is used for data links, satellite handoff channels, drone control frequencies (particularly around 433 MHz, 868 MHz, and 2.4 GHz), and many modern software-defined radio waveforms. The practical SIGINT challenge is that an adversary's relevant emissions span this entire 30 MHz--3 GHz window, which demands DF hardware capable of operating across multiple sub-bands with consistent bearing accuracy.

Operationally, tactical DF is deployed in two distinct modes. Static DF uses fixed or semi-fixed sites -- hilltop installations, forward operating base perimeters, or elevated observation points -- to provide continuous coverage of a defined area with the highest achievable bearing accuracy. Mobile DF deploys on vehicles, small boats, or dismounted teams that maneuver to achieve favorable geometry against a specific emitter or to respond to a collection tasking. The distinction matters for antenna array design and algorithm selection: static sites can support large, carefully calibrated arrays with many elements, while mobile platforms require compact arrays that tolerate the vibration, mutual-coupling, and platform-motion effects of a moving vehicle. Most tactical SIGINT architectures combine both modes, with static sites providing persistent coverage and mobile platforms cued to prosecute specific emitters that the static network has detected.

The frequency agility of modern tactical radios -- frequency-hopping waveforms that change channel every few milliseconds -- forces the DF system to make bearing estimates from very short signal captures, sometimes as brief as 5--10 ms per hop. This constrains algorithm selection: techniques that require long observation windows to build up sufficient statistics cannot function against frequency-hopping emitters. The operational requirement for instantaneous bearing estimation from short signal snapshots is the central performance driver for tactical VHF/UHF DF hardware and algorithm design.

Antenna array design for mobile VHF/UHF DF platforms

The antenna array is the hardware element that most directly determines a DF system's performance ceiling. No amount of signal processing can recover bearing accuracy that the array geometry and calibration do not support. For vehicle-mounted VHF/UHF DF, the dominant array types are the Adcock array and the circular switched array, each with distinct performance trade-offs that suit different parts of the frequency band.

The Adcock array consists of four vertical dipole or monopole elements arranged at the corners of a square, with a fifth omnidirectional sense element at the center. Pairs of opposing elements form two crossed loops whose output voltages are proportional to the sine and cosine of the bearing angle. The inter-element spacing is typically 0.5--1.0 m, providing a full-aperture baseline of 0.7--1.4 m. This aperture gives useful bearing sensitivity from roughly 30 MHz (where the element spacing is a small fraction of a wavelength) up to about 300 MHz (where half-wavelength spacing is reached and phase aliasing begins to become a concern). For UHF coverage above 300 MHz, the aperture must either be reduced to maintain unambiguous phase relationships -- accepting lower accuracy -- or the system must use a larger array with more elements and an interferometric algorithm that can resolve the resulting phase ambiguities. Many production vehicle-mounted DF systems use a dual-band approach: an Adcock array for the VHF band and a separate small circular array for the UHF band, driven by independent receiver chains.

The circular switched array uses 8 to 16 vertically polarized elements arranged at equal angular spacing on a circle, with electronic switching that sequentially connects each element to the receiver. When switched rapidly -- typically at rates of 10--100 kHz -- the switching creates a synthetic rotation that can be processed either as a Doppler signature (the electronically synthesized rotation imparts a frequency modulation whose phase encodes bearing) or as an interferometric set of instantaneous phase samples. The circular array's primary advantage for mobile platforms is mechanical: the elements are physically small at VHF, the array has rotational symmetry that simplifies calibration, and the absence of a large ground plane requirement makes roof-mount installation straightforward. The switching architecture also enables the array to cover the full VHF/UHF range within a single hardware form factor by adapting the switching rate and element selection to the operating frequency.

Watson-Watt and interferometric DF algorithms: principles and accuracy bounds

The Watson-Watt algorithm is the oldest and most widely deployed DF technique for tactical VHF systems. It processes the outputs of two crossed Adcock pairs -- call them the north-south pair (producing voltage V_NS proportional to cos(theta)) and the east-west pair (V_EW proportional to sin(theta)) -- and computes the bearing as theta = atan2(V_EW, V_NS). The sense antenna resolves the 180-degree ambiguity inherent in the crossed-loop geometry by comparing the phase of the sense output to the loop outputs. Because it requires only a single coherent receiver snapshot per bearing estimate, Watson-Watt is well-suited to frequency-hopping intercept: it produces a bearing estimate from every hop capture, and those estimates can be averaged over multiple hops to reduce noise.

The principal accuracy limitation of Watson-Watt is the signal-to-noise ratio dependence of the atan2 computation. When both V_NS and V_EW are small -- as occurs when the emitter is off the broadside of both loops simultaneously, or at low SNR -- the bearing estimate is dominated by noise rather than signal. Watson-Watt achieves typical bearing accuracies of 3--8 degrees RMS under operational conditions, with performance degrading to 10--20 degrees at SNR below 10 dB. Systematic errors from mutual coupling between the array elements, asymmetry in the sense antenna pattern, and near-field scattering from the vehicle body introduce biases that are removed through full-azimuth calibration but that return if the array mounting geometry changes.

Interferometric DF computes bearing from the phase differences between pairs of array elements with known baseline vectors. For a two-element baseline of length d oriented at angle phi relative to north, the phase difference between the elements is delta_phi = (2*pi*d/lambda) * cos(theta - phi), where theta is the emitter bearing and lambda is the wavelength. With multiple baselines at different orientations, the bearing is estimated by finding the theta that best fits all the observed phase differences -- a maximum likelihood problem that can be solved efficiently by grid search or by iterative Newton-Raphson methods. Interferometric DF achieves bearing accuracies of 1--3 degrees RMS on coherent VHF signals at 20 dB SNR, significantly better than Watson-Watt, but at the cost of phase ambiguity when the element spacing exceeds half a wavelength. Resolving phase ambiguity requires either short baselines (sacrificing accuracy) or a multi-baseline array in which short baselines provide unambiguous coarse estimates that are refined by the longer baselines.

Doppler DF for rapidly rotating platforms and compact antenna systems

Doppler DF exploits the fact that an antenna element moving on a circular path around the incoming wavefront experiences a periodic Doppler frequency shift whose instantaneous value depends on the angle between the direction of motion and the emitter bearing. For a circular motion of radius r at angular rate omega, the instantaneous frequency shift is (r*omega/lambda) * sin(theta - omega*t), where theta is the emitter bearing and t is time. This creates a sinusoidal frequency modulation on the received signal at rate omega, with a phase offset equal to the emitter bearing. The bearing estimate is extracted by demodulating the FM signature and measuring its phase -- a process that is algebraically simple and robust to amplitude variations in the received signal.

For electronically switched circular arrays, the physical rotation is replaced by rapid sequential switching between array elements. The switching sequence is designed to synthesize the same FM signature that physical rotation would produce, without any moving parts. Switching rates of 10--100 kHz are typical, with the rate chosen to place the synthetic Doppler tone well within the audio bandwidth of the receiver's demodulator. The key advantage of electronic Doppler DF over interferometric processing is its tolerance of array imperfections: because the bearing information is encoded in the phase of a tone rather than in precise inter-element phase differences, small errors in element position or phase calibration produce small systematic bearing biases rather than the catastrophic phase-unwrapping failures that interferometric algorithms can suffer when calibration is poor.

The accuracy ceiling for Doppler DF is set by the circular array radius relative to wavelength. A larger radius produces a larger FM deviation index and thus a more precisely measurable tone phase. For a 0.2 m radius array at 150 MHz (wavelength = 2 m), the FM deviation index is 2*pi*0.2/2 = 0.63 radians, which translates to a theoretical bearing accuracy of approximately 3--5 degrees RMS at 20 dB SNR. Increasing the radius to 0.5 m improves this to 1.5--2.5 degrees. Vehicle-mounted Doppler DF systems with arrays in the 0.3--0.8 m radius range achieve 2--5 degrees RMS in practice across the VHF band, sufficient to provide useful bearing lines for network-level geolocation even if single-platform accuracy is too coarse for direct position reporting.

Multipath and urban canyon effects on VHF/UHF bearing accuracy

All DF algorithms assume that the received signal is a single plane wave arriving from the true direction of the emitter. This assumption fails in any environment where reflecting surfaces redirect a copy of the emitter's signal toward the DF array from a different angle. The result is that the array sees a superposition of the direct path and one or more reflected copies, and the DF algorithm reports a bearing that is a weighted combination of all the incoming directions. In open terrain with few large reflectors, multipath is typically limited to a ground-reflected component arriving from below the horizon, which Adcock arrays are inherently insensitive to because they use vertically polarized elements with a nulled response at low elevation. In urban environments or dense forests, reflections arrive from all azimuths at angles within the array's main response region, producing bearing errors of 5--30 degrees that no calibration can remove.

Several algorithmic approaches mitigate multipath in practical deployments. Spatial smoothing -- averaging bearing estimates computed over a sequence of signal snapshots acquired while the platform moves -- exploits the spatial decorrelation of the multipath components: the direct-path signal maintains a consistent bearing as the platform moves, while reflected copies shift bearing as the geometry changes. For a platform moving at 30 km/h, a 5-second averaging window covers 42 m of baseline, sufficient to decorrelate multipath components separated by more than a few wavelengths at VHF. The trade-off is that spatial smoothing is inappropriate for stationary platforms and introduces latency that degrades performance against brief transmissions.

Subspace-based algorithms such as MUSIC (MUltiple SIgnal Classification) and ESPRIT can resolve multiple simultaneous signals arriving from different directions, provided the array has sufficient elements and the signals are sufficiently decorrelated. When multipath components are coherent with the direct path -- as occurs when the reflection path length difference is less than the signal's coherence length -- standard MUSIC fails because the signal subspace collapses to a single dimension regardless of how many arriving wavefronts are present. Spatial smoothing of the covariance matrix across subarrays can restore rank and recover MUSIC's ability to resolve coherent multipath, at the cost of reduced effective aperture. In practice, hybrid TDOA/DF approaches that combine bearing lines with time-difference measurements are more robust to coherent multipath than any single-site DF algorithm.

Mobile DF network architecture: coordinating multiple platforms for fix quality

A single DF platform produces a bearing line: a half-infinite ray from the platform's position in the direction of the estimated bearing. The emitter could be anywhere along that ray from a few kilometres to the radio horizon. Converting bearing lines into position fixes requires at least two platforms, and achieving operationally useful CEP values across a realistic emitter area requires careful attention to platform geometry, data link latency, timing synchronization, and the fusion algorithm that combines the bearing reports.

The geometry of a two-platform DF network determines the fix quality through the crossing angle -- the angle at which the two bearing lines intersect at the emitter. When the crossing angle is 90 degrees and both platforms have equal bearing uncertainty sigma_b, the CEP of the intersection is approximately (sigma_b * R) / sin(90 deg) = sigma_b * R, where R is the average range from the platforms to the emitter. For sigma_b = 3 degrees and R = 15 km, CEP is approximately 800 m. When the crossing angle is only 20 degrees -- as occurs when both platforms are nearly collinear with the emitter -- CEP degrades by a factor of sin(90 deg) / sin(20 deg) = 2.9, yielding 2.3 km CEP from the same bearing quality. This geometric dilution of precision (GDOP) is the primary reason that mobile DF networks must maneuver platforms to achieve favorable angles, not merely maximize range to the emitter.

The data link and timing architecture of a mobile DF network must ensure that bearing reports from different platforms can be correlated to the same transmission event. VHF push-to-talk transmissions may last only 2--10 seconds; frequency-hopping waveforms expose each hop for 5--10 ms. Platform clocks must be synchronized to GPS-disciplined timing with sub-millisecond accuracy so that the fusion node can match bearing reports by timestamp. Bearing report messages should carry the platform position, heading, and velocity at the time of the measurement, along with the signal frequency, estimated bearing, bearing uncertainty, and a signal fingerprint (bandwidth, modulation estimate, or power spectral density snapshot) that enables the fusion node to confirm that multiple platforms intercepted the same emitter rather than different emitters on the same frequency. The architecture of on-node versus centralized SIGINT processing directly governs how much of this correlation logic is distributed to the platform versus handled at the fusion node.

Integration with SIGINT collection tasking and track databases

Tactical DF does not operate in isolation. It is embedded in a SIGINT collection chain that includes tasking authorities (who specify which emitters to prosecute and with what priority), collection sensors (which include DF platforms but also non-DF receivers that capture signal content), and analytical databases that accumulate signal history into emitter tracks. Integrating VHF/UHF DF bearing data into this chain requires that the DF system speak the same data formats, timing conventions, and emitter identification schemes as the rest of the collection infrastructure.

Emitter identification is the process of associating a new intercept with a previously catalogued emitter record. Two transmissions on the same frequency are not necessarily from the same emitter: frequency reuse, relay chains, and spectrum congestion all produce ambiguities. DF bearing consistency is one of the most reliable discriminants -- if two intercepts on the same frequency produce bearing lines that converge on the same geographic point, they are almost certainly from the same emitter. The SIGINT database uses bearing history, along with signal fingerprint similarity, temporal pattern analysis, and operator annotations, to maintain emitter track continuity across gaps in collection coverage. When the DF network relocates -- platforms move, coverage geometry changes -- the track association logic must handle the resulting gaps without splitting a single emitter into multiple tracks or merging distinct emitters into one.

Collection tasking integration means that the DF network's scan priority, dwell time per frequency, and bearing report transmission rate are all dynamically adjusted in response to collection priorities set by the tasking authority. A high-priority emitter that has just appeared on the network triggers increased dwell at its known frequency, repositioning of mobile platforms for better geometry, and real-time bearing report forwarding to the fusion node rather than batched transmission. Lower-priority monitoring tasks run in background, contributing to the emitter track database during periods when no high-priority emitter is active. This priority-driven architecture requires a software interface between the collection management system and the DF platform's receiver scheduler -- an interface that, in modern systems, is implemented as a structured command stream over the same data link used for bearing report transmission, allowing the collection manager to retask DF platforms remotely without human intervention at the platform location.