FMCW Radar Systems Overview

• FMCW radar-based systems are coherent architectures that use continuous linear chirps for precise range, velocity, and angular measurements.
• They achieve sub-centimeter resolution and high Doppler accuracy through efficient FFT-based processing and linear frequency modulation.
• Applications span automotive safety, medical monitoring, industrial imaging, and integrated radar-communications with effective interference mitigation.

A Frequency-Modulated Continuous-Wave (FMCW) radar-based system is a coherent radar architecture distinguished by the use of frequency-modulated (typically linear-chirp) signals for high-precision estimation of target range, velocity, and often angle, using continuous transmission and reception rather than discrete pulses. The unique modulation format, high range and Doppler resolution, favorable real-time processing requirements, and increasingly integrated hardware platforms have established FMCW as the dominant paradigm across automotive, imaging, medical, and industrial sensing domains. This article surveys the physics, signal processing chains, waveform and hardware architectures, interference and network protocols, and state-of-the-art applications of FMCW radar-based systems, with traceable references to recent academic literature, open-source simulators, and benchmark datasets.

1. Physical Principles and Signal Model

FMCW radars transmit a continuous, linearly frequency-modulated (chirp) waveform, typically expressed as

$$s_\mathrm{tx}(t) = A \exp\left[j 2\pi (f_0 t + \frac{S}{2} t^2)\right],$$

where $f_0$ is the start (carrier) frequency, $S=B/T$ is the sweep rate (Hz/s), $B$ is the sweep bandwidth, and $T$ is the chirp duration (Bi et al., 2024, Abdulatif et al., 2017).

A point target at range $R$ imposes a delay $\tau = 2R/c$, mapping into a beat frequency

$f_b = S \tau = \frac{2 S R}{c}.$

After mixing and low-pass filtering (dechirp), the returned echo for each chirp yields a superposition of sinusoids corresponding to targets in different range bins. The range resolution is governed by the sweep bandwidth, $\Delta R = c/(2B)$, and is routinely reported at sub-centimeter scale for GHz-level B (Bi et al., 2024, Bryllert et al., 2023). Velocity (Doppler) extraction exploits periodic phase progression across chirps, yielding $f_D = 2 v f_0/c$, separable from the beat in a 2D Range–Doppler FFT processing pipeline (Bi et al., 2024, Abdulatif et al., 2017).

2. Waveform Engineering, Hardware, and Aperture Design

FMCW system hardware typically comprises a voltage-controlled oscillator, mixers for up/down-conversion, and coherent baseband digitization (Bi et al., 2024, Bryllert et al., 2023). High-frequency (>70 GHz) implementations leverage frequency multipliers and compact, high-gain antenna arrays or quasi-optical elements for spatial selectivity (Hamidi et al., 2024, Bryllert et al., 2023). MIMO and virtual-aperture architectures synthesize large effective arrays from a limited number of TX/RX elements, including cascaded MMICs for large virtual element count (Hamidi, 2022).

Key implemented systems include: | Band | Sweep BW | ΔR | Antenna | Ref | |--------------|----------|------------|-------------------|-------------------| | 94/120 GHz | 6–14 GHz | 1–2.5 cm | Narrow beam (3–11°)| (Abdulatif et al., 2017) | | 77 GHz | 3.6 GHz | 3.7 cm | TIDEP-01012, 4xAWR1243 | (Hamidi, 2022) | | 140 GHz (D) | 60 GHz | 2.5 mm | 8×8 URA | (Korlapati et al., 2023) | | 340 GHz | 30 GHz | ∼1 cm | 100 mm parabola | (Bryllert et al., 2023) |

Multi-purpose radar/communications systems embed information via index modulation (IM), spectral allocation, or joint waveform multiplexing with FMCW, enabling dual-function radar–communications (DFRC) while minimizing hardware overhead (Ma et al., 2021, Gaviria et al., 21 Mar 2025). Synthetic aperture radar (SAR) is realized by mechanical or vehicular motion, or by 2D MIMO array synthesis, to enable sub-cm cross-range imaging along with high range resolution (Hamidi et al., 2024, Hamidi, 2022, Park et al., 2020).

3. Signal Processing Chain and Algorithmic Methods

The canonical FMCW processing chain comprises several stages:

1. Mixing and Filtering: Echo signal is mixed with a local chirp replica and low-pass filtered to produce the intermediate frequency (IF) “beat” signal encoding range (via beat frequency) (Abdulatif et al., 2017, Bi et al., 2024).
2. Fast-Time FFT (Range FFT): IF signal samples per chirp are windowed and FFT’d, yielding a range profile for every chirp. $N$ FFT bins maps to range bins of width $\Delta R$ (Bi et al., 2024, Gao, 8 Sep 2025).
3. Doppler FFT (Slow-Time): Stack sequential chirp FFTs, then FFT across chirps to measure target velocity (Doppler bins) (Bi et al., 2024, Abdulatif et al., 2017, Hadjipanayi et al., 7 Jan 2026).
4. Angle-of-Arrival Processing: Implemented via spatial FFT, MUSIC/MVDR, or compressive sensing methods across TX/RX arrays for angular or 3D localization (Korlapati et al., 2023, Gao et al., 2021).
5. Micro-motion and Vital Sign Estimation: High-resolution phase tracking extracts sub-mm periodic breathing/heartbeat displacements in medical monitoring (Benny et al., 26 Nov 2025, Abdulatif et al., 2017).
6. SAR and Imaging Algorithms: Range-compressed data is processed via migration/backprojection or w–k transforms for 2D/3D imaging (Hamidi et al., 2024, Hamidi, 2022, Park et al., 2020).
7. Activity Recognition and Gait Analysis: Extract range–Doppler–time cubes and leverage CNN, LSTM, or transformer architectures for HAR/gait from spectrograms (Bi et al., 2024, Gao, 8 Sep 2025, Hadjipanayi et al., 7 Jan 2026).

Digital pipeline implementations routinely leverage NumPy/SciPy (Python), MATLAB, or C++ for dataflow, FFTs, and matrix operations (Bi et al., 2024). Modern ML-based systems operate directly on range–Doppler maps or point clouds for classification and object tracking (Gao, 8 Sep 2025, Bi et al., 2024).

4. Interference, Network Coordination, and Mitigation

FMCW radars experience mutual interference and spoofing due to spectral overlap, especially in dense networks (e.g., autonomous vehicles). Addressed schemes include:

• Random-frequency hopping (BlueFMCW), in which chirps are split into randomized sub-chirps to whiten interferer/spoofer energy across the beat spectrum, preserving full range resolution after phase alignment and realignment algorithms (Moon et al., 2020).
• Medium access protocols for radar networks, e.g., slotted ALOHA and CSMA with clear-channel assessment (CCA), in which radar transmissions are scheduled to minimize narrowband ghost-target interference. Notably, maximizing probing (p→1) in CSMA often maximizes throughput and robustness unlike standard communication networks (K et al., 2022).
• Matrix-pencil and CFAR-based interference mitigation: Short contaminated segments in the time domain are located and excised, then reconstructed via sparse parameter estimation (matrix-pencil), or via 1D cell-averaging CFAR detection in the time–frequency plane followed by zeroing or amplitude correction (wang et al., 2021, wang, 2021). These approaches restore target detectability and reduce power loss without impairing the desired signal.

5. Imaging, Medical, and Industrial Applications

FMCW radar-based systems enable advanced perception and monitoring applications:

• All-weather, high-resolution mmWave imaging: W-band (75–110 GHz) and sub-mm (340 GHz) systems achieve sub-cm and mm-level range/cross-range resolution for near-field SAR, industrial particle tracking, and automotive perception under adverse conditions (rain, snow, fog) (Hamidi et al., 2024, Bryllert et al., 2023, Gao et al., 2021).
• Medical vital-sign and multi-patient monitoring: 77–120 GHz FMCW radars track sub-cm chest micro-motions (breathing, heartbeats), with array-based digital beamforming for multi-person detection and >97%/93% breath/heart rate accuracy (Benny et al., 26 Nov 2025, Abdulatif et al., 2017).
• Human activity recognition/gait analysis: Processing via range–Doppler and micro-Doppler time–frequency representations, followed by deep learning, enables robust in-home behavioral monitoring, with radar-agnostic frameworks validated equally against IR-UWB and FMCW (Gao, 8 Sep 2025, Hadjipanayi et al., 7 Jan 2026).
• Scene mapping and height estimation: Multipath-induced amplitude modulation, extractable via post-FFT processing, enables estimation of object height (e.g., vehicles vs. trucks) robustly with standard automotive FMCW (Kohnert et al., 2023).

Industrial applications extend to mm-level material characterization, non-contact flow/particle monitoring, and through-wall sensing (Bryllert et al., 2023, Hamidi, 2022, Hamidi et al., 2024).

6. Integrated Sensing and Communications (ISAC)

FMCW radar systems are increasingly leveraged for integrated sensing and communications. Arbitrary (high-order) constellation modulation (QAM/PSK), when overlaid on the base FMCW, can achieve high data rates without sacrificing sensing accuracy or increasing SNR penalty above 0.2 dB relative to pure FMCW, due to precise signal-alignment and compensation steps (RVPC, data division), and careful waveform design (Gaviria et al., 21 Mar 2025). DFRC platforms exploit index modulation (spatial and spectral sub-carrier selection) with sparse arrays to maximize bitrates and spatial resolution without increased front-end complexity (Ma et al., 2021).

Key findings are:

Approach	Comms Rate	Sensing Penalty	Comments
FMCW+QAM/PSK	~384 kbit/s/chirp	<0.2 dB	Arbitrary constellations possible
IM/DFRC (FRaC)	5–9 bit/pulse	~1 dB at BER	Sparse transmit, virtual MIMO

7. Limitations, Trade-offs, and Future Directions

While FMCW systems enable high-resolution and cost-efficient sensing, several practical trade-offs must be addressed:

• Mutual interference remains substantial in dense, uncoordinated deployments. Randomization (BlueFMCW), scheduling (CSMA), and digital post-processing (matrix-pencil, CFAR-AC) each offer partial mitigation (Moon et al., 2020, K et al., 2022, wang, 2021, wang et al., 2021).
• Bandwidth constraints: Ultra-wide sweeps enable <1 cm resolution, but face regulatory and hardware challenges.
• Array size/hardware complexity: MIMO and virtual arrays require careful RF distribution network, A/D selection, and mitigation of mutual coupling—especially at D-band and sub-mm frequencies (Korlapati et al., 2023, Hamidi, 2022).
• Calibration and real-time requirements: Motion compensation, synchronization, and online array calibration are mandatory for mm-precision imaging and multisubject medical applications (Hamidi, 2022, Benny et al., 26 Nov 2025).
• SAR and imaging: High-resolution imaging is fundamentally limited by synthetic aperture length (cross-range), requiring either mechanical motion or large, dense MIMO arrays (Hamidi et al., 2024).
• ML robustness and domain adaptation: For HAR, the performance of deep models is currently state-of-the-art, but cross-domain transfer and radar-agnostic fusion pipelines remain active research areas (Bi et al., 2024, Gao, 8 Sep 2025, Hadjipanayi et al., 7 Jan 2026).

Long-term developments include real-time 3D MIMO arrays, edge-optimized deep learning, adaptive waveform design for interference avoidance, and integrated fusion across sensor modalities for robust autonomous and industrial systems.

---

References:

• (Bi et al., 2024) FMCW Radar Principles and Human Activity Recognition Systems: Foundations, Techniques, and Applications
• (Abdulatif et al., 2017) Power-Based Real-Time Respiration Monitoring Using FMCW Radar
• (Hamidi et al., 2024) High Resolution Millimeter Wave Imaging Based on FMCW Radar Systems at W-Band
• (Hamidi, 2022) 3D Near-Field Virtual MIMO-SAR Imaging Using FMCW Radar Systems at 77 GHz
• (Korlapati et al., 2023) D-Band 2D MIMO FMCW Radar System Design for Indoor Wireless Sensing
• (Kohnert et al., 2023) FMCW Radar Height Estimation of Moving Vehicles by Analyzing Multipath Reflections
• (Benny et al., 26 Nov 2025) Design and Measurements of mmWave FMCW Radar Based Non-Contact Multi-Patient Heart Rate and Breath Rate Monitoring System
• (Hadjipanayi et al., 7 Jan 2026) Towards Radar-Agnostic Gait Analysis Across UWB and FMCW Systems
• (Ma et al., 2021) FRaC: FMCW-Based Joint Radar-Communications System via Index Modulation
• (Gao, 8 Sep 2025) RadHARSimulator V1: Model-Based FMCW Radar Human Activity Recognition Simulator
• (Gaviria et al., 21 Mar 2025) On the Sensing Performance of FMCW-based Integrated Sensing and Communications with Arbitrary Constellations
• (Moon et al., 2020) BlueFMCW: Random Frequency Hopping Radar for Mitigation of Interference and Spoofing
• (K et al., 2022) Slotted ALOHA and CSMA Protocols for FMCW Radar Networks
• (wang et al., 2021) Matrix-Pencil Approach-Based Interference Mitigation for FMCW Radar Systems
• (wang, 2021) CFAR-Based Interference Mitigation for FMCW Automotive Radar Systems
• (Park et al., 2020) FMCW SAR with New Synthesis Method Based on A-SPC Technique
• (Bryllert et al., 2023) A Submillimeter-Wave FMCW Pulse-Doppler Radar to Characterize the Dynamics of Particle Clouds
• (Gao et al., 2021) Perception Through 2D-MIMO FMCW Automotive Radar Under Adverse Weather