Satellite-Illuminated Passive Bistatic Radar for Low-Altitude Aerial Object Detection Using Commodity Ku-Band Hardware

Author: River Caudle
Affiliation:
Status: Defensive Publication
Date: May 2026
License: CC0 / Public Domain Dedication

Abstract

This document describes a system architecture for persistent overhead detection of low-altitude aerial objects (drones, UAVs, birds, balloons) using geostationary and low-earth-orbit satellite downlinks as illuminators of opportunity. The system uses an array of consumer-grade satellite television LNBs (Low Noise Block downconverters) modified for phase coherence, a commodity 5-channel coherent software-defined radio as the IF backend, and digital signal processing techniques borrowed from established passive bistatic radar literature. Total hardware cost is estimated under USD 1500 in parts. The system requires no RF transmission, no spectrum allocation, no operating license, and produces no detectable emission signature. This document is published to establish prior art and to provide a reference architecture for open implementation.

1. Problem Statement

Existing commercial counter-UAS (anti-drone) systems fall into three categories: active radar, RF emission detection, and electro-optical sensors. Each has limitations.

Active radar systems require licensed transmit operation, draw attention through their own emissions, and are expensive (typically USD 50,000 to USD 500,000 per site). RF emission detection systems are defeated by RF-silent autonomous drones operating on visual navigation. Electro-optical sensors are weather-limited and have poor performance against small targets at night.

There is an unfilled gap for a low-cost, fully passive, weather-resistant, persistent overhead surveillance modality that can detect small slow-moving aerial objects without itself being detectable.

This document proposes such a system based on satellite-illuminated passive bistatic radar (SIPBR or satellite PBR).

2. Conceptual Overview

A geostationary satellite at 35,786 km altitude continuously illuminates a wide footprint of the earth's surface with high-power radio frequency signals. Direct-to-home (DTH) television satellites in Ku-band (10.7 to 12.75 GHz) provide effective isotropic radiated power (EIRP) of typically 50 to 55 dBW at the edge of beam, with continuous coverage 24 hours per day. Similar signals are available in C-band (3.7 to 4.2 GHz) and Ka-band (18 to 20 GHz) from numerous other satellites.

Any aerial object passing through the illuminated airspace between a satellite and a ground receiver acts as a passive reflector. It re-radiates a small fraction of the incident energy in all directions, including back toward the ground. By comparing the direct-path signal (received on a reference channel) with the reflected signal (received on one or more surveillance channels), it is possible to extract a Doppler-shifted, time-delayed echo whose properties (Doppler shift, bistatic range, angle of arrival) uniquely characterize the aerial target.

This is a textbook application of passive bistatic radar (radar passif bistatique), well-established in the academic literature for aircraft detection using FM broadcast and DVB-T illuminators. The novelty of the present work is the combination of three elements:

1. Use of Ku-band satellite illuminators specifically, which provide higher EIRP, wider bandwidth, and continuous availability compared to terrestrial broadcast sources.
2. Use of modified consumer LNBs as the coherent front end, reducing front-end cost by approximately two orders of magnitude compared to laboratory-grade microwave receivers.
3. Use of a commodity 5-channel coherent software-defined radio (the KrakenSDR or equivalent) as the IF backend, providing turn-key phase coherence and direction-finding capability at the IF stage.

To the author's knowledge, this specific combination has not been deployed commercially nor published as an integrated reference design for drone detection.

3. System Architecture

3.1 RF Front End

The RF front end consists of N coherent Ku-band downconverter channels, where N is between 3 and 5 depending on desired angle-of-arrival accuracy.

Each channel comprises:

• One satellite dish (parabolic reflector), 60 to 90 cm diameter, pointed at a target patch of sky. Dishes may be pointed at the same GEO satellite (providing a beamformed aperture) or at different satellites (providing independent illumination geometries).
• One Ku-band LNB modified for external local oscillator (LO) injection. Standard LNBs contain an internal dielectric resonator oscillator (DRO) operating at 9.75 GHz or 10.6 GHz. The modification consists of disabling the internal DRO and injecting an external LO from a shared source.
• Coaxial cable to the IF backend, carrying both the IF signal (950 to 2150 MHz) and DC bias to the LNB.

The shared LO source is critical to system function. All N channels must be driven by an LO that is phase-locked across channels, with matched cable lengths or with measured and calibrated phase offsets. Suitable LO sources include the ADF5355 PLL synthesizer with active fanout, or a phase-locked Gunn oscillator distributed via a Wilkinson 1:N splitter.

The LO frequency determines the RF coverage:

• LO at 9.75 GHz: RF coverage 10.7 to 12.1 GHz, IF 950 to 2350 MHz (standard low-band Ku)
• LO at 10.6 GHz: RF coverage 11.7 to 12.75 GHz, IF 1100 to 2150 MHz (standard high-band Ku)

3.2 IF Backend

The IF backend receives the N coherent IF signals and performs ADC conversion, phase calibration, and DSP. A commodity 5-channel coherent SDR such as the KrakenSDR is well-suited to this role:

• Tuning range 24 MHz to 1766 MHz, covers most of the IF band
• Built-in phase coherence between channels via shared 28.8 MHz reference clock
• Built-in noise-source calibration hardware for IF-side phase calibration
• Open-source DAQ firmware (heimdalldaqfw) and DSP code (krakensdrdoa, krakensdrpr)
• Cost approximately USD 600

Note that the KrakenSDR's built-in noise calibration corrects for IF-side mismatches only. RF-side mismatches (LNB-to-LNB gain, mixer phase, LO distribution imbalance) require separate calibration. A secondary noise source switchable into all N RF channels simultaneously, via directional couplers ahead of the LNBs, addresses this.

3.3 Reference Channel

One of the N channels (or an additional N+1th channel) is configured as a reference channel, dedicated to capturing the direct-path satellite signal at high signal-to-noise ratio. The reference channel feeds the matched-filter correlator that processes the surveillance channels.

In the simplest configuration, one of the survey channels is also used as the reference, accepting some performance degradation. In more capable configurations, a dedicated reference dish points directly at the strongest visible satellite, while the surveillance dishes are arranged for spatial coverage of the airspace of interest.

3.4 Antenna Array Geometry

For pure detection (presence or absence of a target), a single surveillance dish suffices. For angle-of-arrival estimation, an array of three or more spatially separated dishes is required. For full 3D localization, the array must span enough baseline to provide useful parallax against the target altitude of interest.

For a target zone of approximately 1 km diameter at 100 to 300 m altitude (typical drone threat envelope for facility protection), baselines of 5 to 50 m between dishes are appropriate. Smaller baselines reduce localization accuracy; larger baselines require more careful phase calibration over longer cable runs.

A practical configuration for facility protection:

• One reference dish, fixed elevation, pointed at strongest visible GEO satellite
• Four surveillance dishes arranged in a roughly square pattern, 10 to 20 m apart, pointed slightly off-zenith to maximize forward-scatter geometry for low-altitude targets

4. Physics and Link Budget

4.1 Direct Path

A typical Ku-band DTH satellite delivers approximately -110 dBW/m² at the ground in its primary footprint. A 60 cm dish has approximately 35 dBi gain at 12 GHz, capturing roughly -75 dBW (-45 dBm) of signal power in a 30 MHz transponder bandwidth. This is approximately 60 dB above thermal noise floor, an extremely strong signal by RF survey standards.

4.2 Reflected Path

The bistatic radar equation gives the received reflected power:

Pr = (Pt × Gt × Gr × λ² × σb) / ((4π)³ × R1² × R_2²)

where:

• Pt × Gt = satellite EIRP, approximately 50 dBW
• G_r = receiver dish gain, 35 dBi
• λ = wavelength at 12 GHz, 0.025 m
• σ_b = bistatic radar cross-section of target, estimated 0.001 to 0.1 m² for a small quadcopter at Ku-band depending on aspect and geometry
• R_1 = satellite-to-target distance, 35,786 km
• R_2 = target-to-receiver distance, 100 to 500 m for low-altitude facility protection

For a quadcopter at 200 m slant range with σ_b = 0.01 m², the received reflected power is approximately -160 dBm. This is approximately 40 dB below thermal noise floor in the full transponder bandwidth, but is recoverable via coherent matched-filter processing against the known reference waveform.

4.3 Forward Scatter Enhancement

When the target is positioned approximately on the line between satellite and receiver, the bistatic RCS approaches the forward-scatter RCS (FSCS), which for targets larger than the wavelength can be 10 to 100 times the monostatic RCS. For a 30 cm wingspan quadcopter at 12 GHz (λ = 2.5 cm), the target is approximately 12 wavelengths across, in the optical regime, and forward scatter gain is substantial.

For drones positioned in the forward-scatter geometry between satellite and receiver, σ_b may be 0.1 to 1 m², increasing received power by 10 to 20 dB. This is the geometry to design for if maximum detection range is the goal.

4.4 Doppler Resolution

Coherent integration time T_int determines Doppler resolution:

Δf = 1 / T_int

For T_int = 1 second, Δf = 1 Hz, corresponding to a radial velocity resolution of approximately 12.5 mm/s at 12 GHz. This is more than adequate to discriminate a hovering drone (sub-meter/second motion from prop wash and station-keeping) from stationary clutter.

For T_int = 10 seconds, Δf = 0.1 Hz, sufficient to detect a perfectly hovering drone via micro-Doppler from rotor blade modulation alone.

Long integration times require correspondingly stable phase across the system, which is the central engineering challenge addressed by the modified-LNB coherent architecture.

4.5 Range Resolution

A DVB-S transponder carries 27 to 36 MHz of bandwidth. Matched-filter range resolution is c / (2 × B) ≈ 4 to 6 m in the bistatic sense. This is adequate for distinguishing a drone from background clutter at the ranges of interest.

Higher resolution is achievable by using multiple transponders simultaneously (the KrakenSDR's per-channel bandwidth limit of 2.56 MHz is the practical bottleneck here, not the satellite signal). Using a higher-bandwidth backend (USRP B210 at 56 MHz, X310 at 200 MHz) substantially improves range resolution at corresponding cost increase.

5. Digital Signal Processing Architecture

The processing pipeline implements established passive bistatic radar techniques:

1. Reference channel acquisition. Demodulate and re-encode the direct-path signal to obtain a clean, noise-free copy of the transmitted waveform.

1. Direct path interference (DPI) cancellation. The direct path also leaks into the surveillance channels via antenna sidelobes and is dramatically stronger than the desired echo. Adaptive cancellation using algorithms such as ECA (Extensive Cancellation Algorithm) or LMS-based projection removes the direct path while preserving the echo. This is the most computationally demanding stage and is the central technical challenge.

1. Range-Doppler matched filtering. Cross-correlate the cancelled surveillance signal with time-delayed and Doppler-shifted copies of the reference. This produces a range-Doppler map.

1. Constant false alarm rate (CFAR) detection. Apply adaptive thresholding to the range-Doppler map to identify target detections above clutter and noise.

1. Angle of arrival estimation. For each detection, use phase relationships across the N surveillance channels to estimate the angle of arrival. Standard array processing algorithms (MUSIC, ESPRIT, Capon beamformer) apply directly.

1. Tracking. A Kalman filter or similar state estimator maintains tracks across consecutive detections, providing trajectory estimates and supporting classification.

1. Classification. Doppler signature (including micro-Doppler from rotor blades) and trajectory characteristics support classification (drone vs bird vs aircraft vs balloon). Machine learning classifiers trained on known-target signature databases improve performance.

Open-source building blocks exist for most of these stages. The integration is the work.

6. Implementation Notes

6.1 Bill of Materials (5-channel reference design)

| Item | Qty | Unit Cost | Total |
|------|-----|-----------|-------|
| Ku-band offset dish, 60 cm | 5 | $50 | $250 |
| Ku-band LNB, single-output | 5 | $20 | $100 |
| LO source (ADF5355 eval board) | 1 | $150 | $150 |
| LO distribution amplifier and splitter | 1 | $100 | $100 |
| GPSDO 10 MHz reference | 1 | $200 | $200 |
| KrakenSDR | 1 | $600 | $600 |
| Coaxial cable, connectors, bias tees | - | - | $100 |
| Calibration noise source, switch, couplers | 1 | $150 | $150 |
| Compute (Raspberry Pi 5 or small PC) | 1 | $100 | $100 |
| Enclosure, mounting hardware | - | - | $100 |
| Total | | | $1,850 |

A cost-reduced single-channel detection-only variant is achievable for under $400 (one dish, one LNB, one RTL-SDR, one Pi).

6.2 LNB Modification

The LNB modification has been documented by amateur radio operators for radio astronomy purposes (notably for 1420 MHz hydrogen line observation using arrays of phase-locked LNBs). The general procedure:

1. Open the LNB enclosure.
2. Locate the dielectric resonator (small ceramic puck adjacent to the FET mixer transistor).
3. Disable the internal DRO either by removing the resonator entirely or by applying a small conductive patch to detune it.
4. Identify the LO injection point in the mixer circuit. This is typically the gate or drain of the mixer FET, depending on LNB design.
5. AC-couple the external LO via a small capacitor (10 pF) to the injection point.
6. Verify mixer operation with a known test signal.

This modification is straightforward for someone comfortable with SMD rework. It is irreversible without spare LNBs, so practice on cheap units first.

Phase noise of the resulting system is determined by the external LO, not by the LNB internals. A clean external LO (typical phase noise -100 dBc/Hz at 10 kHz offset) produces substantially better phase coherence than the stock DRO of a consumer LNB.

6.3 Calibration Procedure

System calibration occurs in two stages:

• IF-side calibration is handled automatically by the KrakenSDR's built-in noise source, correcting for phase drift between the five RTL-SDR receivers.
• RF-side calibration must be added externally. A noise source (Mini-Circuits NSAS-15+ or similar) is switched into all five surveillance channels simultaneously through a 1:5 power splitter and directional couplers ahead of each LNB. Periodic injection (every few minutes) followed by phase measurement allows the system to track and correct for thermal drift in the LNBs and LO distribution.

Initial geometry calibration establishes the relative phase offsets between channels at known frequencies and antenna positions. This is a one-time process that produces a calibration table loaded at startup.

6.4 Operating Considerations

• Weather. Rain attenuation at Ku-band (rain fade) reduces direct-path SNR and degrades detection performance. The system remains functional but with reduced sensitivity during heavy precipitation. C-band variants are more weather-robust.
• Satellite handover. GEO satellite illuminators are continuously available. No handover or scheduling required.
• Bandwidth allocation. The KrakenSDR's 2.56 MHz per channel limits the effective bandwidth used from any single transponder. Higher-bandwidth backends improve range resolution.
• Multi-satellite operation. A single dish receives one satellite at a time. Multi-feed dishes or multiple dishes provide simultaneous coverage of multiple satellites, improving geometric diversity and detection robustness.

7. Limitations and Open Problems

7.1 Known Limitations

• Coverage geometry. The system detects targets in the illuminated volume between satellite and receiver. Full sky coverage requires multiple satellite illuminators or multiple receiver sites.
• Slow targets. Pure detection of stationary or near-stationary targets requires very long coherent integration and exceptional phase stability. Hovering drones are detectable via rotor micro-Doppler, but a stationary balloon is essentially invisible.
• Range ambiguity. Bistatic range is not the same as monostatic range. Converting bistatic detections to 3D positions requires geometric solution and is non-trivial.
• Calibration sensitivity. Performance degrades rapidly with phase calibration error. Thermal management of the RF front end is essential for operational stability.
• Computation. Real-time direct-path cancellation and range-Doppler matched filtering at Ku-band bandwidths is computationally demanding. A modest workstation or Jetson-class compute module is required.

7.2 Open Problems

• Classification of low-RCS targets. Distinguishing small drones from large birds via Doppler signature is non-trivial. Training data sets are not publicly available.
• Multi-target tracking in cluttered environments. Performance in urban environments with many moving reflectors (vehicles, pedestrians, swaying vegetation) is unknown.
• Optimal antenna array geometry. Theoretical work on array geometry for this specific bistatic radar configuration is sparse.
• Real-time DSP implementation. Open-source passive radar codes (PASSIM, Spyserver-based projects) require adaptation to this specific architecture.

8. Variants and Extensions

8.1 Starlink as Illuminator

Starlink satellites in low earth orbit (550 km altitude) transmit Ku-band downlinks with substantially higher received power than GEO satellites (approximately 30 dB stronger due to reduced path loss). Constellation geometry provides multiple simultaneous illuminators visible from any ground site, sweeping across the sky on minute timescales.

A Starlink-illuminated variant offers:

• Higher signal-to-noise ratio
• Wider effective bandwidth (Starlink uses 250 MHz channels)
• Geometric diversity from constellation motion
• Multi-baseline solutions from simultaneous multi-satellite illumination

The disadvantage is greater system complexity, since the illuminator geometry changes constantly and must be tracked. Reference signal acquisition is also more demanding due to Starlink's proprietary waveform.

8.2 C-band and Ka-band Variants

The same architecture applies at other bands:

• C-band (3.7 to 4.2 GHz) using modified C-band LNBs. Better rain robustness, lower resolution, less RCS contrast for small targets.
• Ka-band (18 to 20 GHz) using modified Ka-band LNBs. Higher resolution, more RCS for small drones (wavelength approaches drone feature size), much worse rain attenuation.

A multi-band system using all three offers complementary capabilities.

8.3 Mobile Deployment

The reference architecture is fixed-site. A mobile (vehicle-mounted) variant is mechanically feasible but introduces additional challenges: motion compensation for the receiver platform, dynamic satellite tracking, and GPS-disciplined position estimation. Useful for tactical deployment but a significant additional engineering effort.

8.4 Sensor Fusion

Integration with complementary modalities improves robustness:

• Acoustic detection arrays for short-range confirmation
• Visible-light cameras for visual identification of declared tracks
• Magnetometer arrays for very-low-altitude detection
• Existing ADS-B receivers for cooperative aircraft track filtering

A fused system using passive bistatic radar as the primary detection layer and other modalities for confirmation and classification is the operationally robust configuration.

9. Prior Art

The following bodies of work are relevant:

• Passive bistatic radar literature. Extensive academic work on FM and DVB-T illuminators dating to the 1990s. Key references include Howland (1999), Griffiths and Baker (2005), and the IET Radar, Sonar, and Navigation passive radar series.
• DVB-S passive radar research. Papers from groups at University of Pisa, Warsaw University of Technology, and others on satellite-illuminated passive radar, primarily for maritime and large-aircraft detection.
• GNSS reflectometry. Operational technique for ocean surface monitoring (NASA CYGNSS mission and others). Demonstrates feasibility of using satellite navigation signals for bistatic sensing.
• Coherent LNB arrays for radio astronomy. Amateur and academic projects using phase-locked LNBs for 1420 MHz hydrogen line observation and other passive radio astronomy.
• KrakenSDR direction-finding software. Open-source DOA estimation and passive radar code from KrakenRF.

To the author's knowledge, no published work integrates all of these elements specifically for small drone detection using consumer Ku-band hardware. This document is published as prior art to support open implementation and to forestall patent claims on the obvious integration.

10. Defensive Publication Notice

This document is released into the public domain under CC0 / Public Domain Dedication. The author makes no patent claims on the described system, its components, or any obvious variants. Any subsequent patent claims on the described architecture should be considered prior-art-invalid based on the publication date of this document.

The intent is to keep this technique freely available for use by:

• Open-source defense and security projects
• Academic research groups
• Civil infrastructure protection
• Critical facility operators in jurisdictions without commercial counter-UAS access

The components and techniques are commodity, the integration is straightforward for a competent RF engineer, and there is no legitimate intellectual property claim available to anyone reading this document.

11. Acknowledgments

This architecture synthesizes work from the KrakenRF team (KrakenSDR hardware and DAQ firmware), the amateur radio astronomy community (coherent LNB array techniques), and the academic passive bistatic radar community (signal processing methods). None of the constituent ideas are novel; only the specific integration for low-cost small-drone detection is claimed as a contribution.

12. Contact

River Risk Partners
[contact information]

End of document.