Coherent Integration for Multi-Node Millimeter-Wave Radar

Inspiration

Indoor millimeter-wave radar is becoming an important sensing technology for smart homes, healthcare monitoring, human presence detection, and privacy-preserving perception. A single radar node, however, is limited by its physical aperture, angular resolution, and sensitivity to occlusion and weak reflections.

Our project was inspired by a simple question: if several low-cost 60 GHz radar nodes are mounted on the same wall and observe the same indoor scene, can they be coherently integrated to behave like a much larger virtual aperture?

Instead of treating each radar independently, we aim to align their phase, geometry, timing, and waveform models so that their complex signals can add constructively at true target locations and suppress false responses elsewhere.

What it does

This project designs and simulates a coherent integration framework for multiple wall-mounted millimeter-wave radar nodes.

The system focuses on:

• Multi-node indoor 60 GHz FMCW radar sensing
• Same-wall, same-direction radar deployment
• Distributed aperture formation
• Coherent and non-coherent fusion comparison
• Cross-range resolution improvement
• Weak target enhancement
• Sensitivity analysis for phase, position, timing, and frequency errors
• Indoor multipath modeling using wall-reflection image targets

The baseline scenario uses three TI 60 GHz radar-like nodes mounted on one wall and facing into a room. The simulation compares single-radar imaging, non-coherent fusion, and coherent fusion under controlled target layouts.

How we built it

We first defined a clean indoor simulation environment:

• Room size: 6 m by 5 m
• Radar wall: y = 0
• Radar looking direction: +y
• Radar positions: (1.5, 0, 1.2), (3.0, 0, 1.2), (4.5, 0, 1.2)
• Initial imaging plane: z = 1.2 m

We then designed several simulation scenes:

• Single target calibration
• Two closely spaced targets for resolution testing
• Depth sweep to study range-dependent cross-range resolution
• Multipath scenes using image targets from room walls

For the radar model, we used TI 60 GHz mmWave radar concepts, including FMCW chirp configuration, TDM-MIMO structure, virtual antenna geometry, and near-field phase modeling. A key modeling decision was to compute phase using the exact TX-target-RX path instead of relying only on far-field virtual-array approximations.

The coherent imaging model compensates each radar signal according to the candidate target position. When the candidate position is correct, the phases align and produce a strong coherent peak.

Challenges we ran into

The biggest challenge is that coherent integration is extremely sensitive to small errors.

At 60 GHz, the wavelength is only about 5 mm. This means that millimeter-level geometry errors can cause large phase errors. Position calibration, inter-radar synchronization, carrier phase alignment, and channel phase compensation all become critical.

Another challenge is the difference between standard radar angle processing and near-field coherent fusion. Virtual antenna models are useful for angle FFT, but indoor distributed coherent fusion requires more careful phase-path modeling.

Indoor multipath also complicates the problem. Reflections from walls, floors, ceilings, and furniture can create false targets or interfere with coherent focusing.

Accomplishments that we're proud of

We established a clear simulation roadmap for studying coherent integration in indoor multi-node radar systems.

Key accomplishments include:

• A complete wall-mounted radar deployment model
• Multiple benchmark scenarios for coherent fusion evaluation
• A TI 60 GHz radar-inspired waveform and antenna configuration summary
• A near-field phase model suitable for coherent fusion
• A structured error-sensitivity plan
• Visual scenario diagrams for communicating the system design
• A compact project archive that can be shared with other LLMs or collaborators

The most important accomplishment is narrowing the problem from a vague “multi-radar fusion” topic into a concrete and testable distributed coherent aperture problem.

What we learned

We learned that coherent radar fusion is not just a signal-processing problem. It is also a calibration, geometry, hardware synchronization, and propagation-modeling problem.

Important lessons include:

• Same-wall radar deployment should be treated as distributed aperture synthesis.
• Coherent fusion can improve resolution only if phase alignment is accurate.
• Near-field indoor sensing requires exact path-length modeling.
• TI-style waveform parameters directly affect range resolution, velocity ambiguity, and frame timing.
• Multipath must be introduced gradually after the ideal coherent model works.
• Comparing coherent fusion against non-coherent fusion is essential to prove the value of phase alignment.

What's next for Coherent Integration for Multi-Node Millimeter-Wave Radar

The next step is to implement the full simulation pipeline.

Planned work includes:

• Build the point-target FMCW signal simulator
• Add physical TX/RX antenna phase centers for each radar
• Implement 2D coherent imaging over the room grid
• Compare single-radar, non-coherent, and coherent fusion results
• Sweep target spacing to quantify resolution improvement
• Add phase, timing, position, and frequency errors
• Introduce wall multipath and reflection coefficients
• Extend from point targets to human-body scattering models
• Validate the simulation using real TI 60 GHz radar data if available

The long-term goal is to develop a robust coherent integration method that allows low-cost indoor radar nodes to work together as a calibrated distributed sensing system.

Built With

• c
• iwr6843
• mmwavesdk
• ti