Radar Goes Personal
Radar has protected warships and guided aircraft for over 80 years. Only in the last decade has semiconductor miniaturization made it possible to pack a complete radar transceiver into a module the size of a postage stamp, small enough to mount on a helmet or embed in a vest panel. These millimeter-wave modules operate in the 60-77 GHz frequency bands, where wavelengths of 4-5 millimeters enable antenna arrays compact enough for personal wearable applications while providing angular resolution sufficient to distinguish individual humans at ranges up to 100 meters.
For personal defense systems, radar offers capabilities that no optical sensor can match. Radar operates in total darkness, through smoke, fog, and dust, and most critically, through walls and other solid barriers. A helmet-mounted radar module can detect a person moving behind a standard interior wall at distances up to 15 meters, providing situational awareness that is impossible with cameras or thermal imagers alone.
FMCW Radar Fundamentals
Frequency-Modulated Continuous Wave (FMCW) radar is the architecture of choice for wearable applications. Unlike pulsed radar, which requires high peak power and large antennas, FMCW transmits a continuous signal whose frequency ramps linearly over time, called a chirp. The reflected signal from a target arrives back at the receiver with a time delay proportional to the target's range. By mixing the received signal with the transmitted chirp, the system produces a beat frequency that directly encodes the target distance.
Texas Instruments' IWR6843 and IWR1843 radar-on-chip modules exemplify the state of the art. The IWR6843 integrates three transmit and four receive antennas, a 77 GHz RF front-end, an ARM Cortex-R4F processor, and a C674x DSP core into a single package measuring 10.4 x 10.4 mm. Operating from 60 to 64 GHz, it achieves range resolution of approximately 4 cm and angular resolution of 15 degrees with its integrated antenna array. Power consumption is under 2 watts in active mode.
Through-Wall Detection
Through-wall radar (TWR) uses lower frequencies, typically 1-10 GHz in the UWB range, where electromagnetic waves penetrate building materials more effectively. Standard drywall on wood studs attenuates signals by 3-5 dB at 3 GHz, while concrete block walls produce 10-15 dB of attenuation. Even through concrete, sufficient signal returns to detect human movement at ranges up to 8-10 meters behind the wall.
The physics of through-wall detection create unique processing challenges. The wall itself produces a strong reflection that can mask the much weaker returns from targets behind it. Clutter removal algorithms subtract the static wall return, revealing only the time-varying components caused by human movement. Breathing detection is possible through micro-motion analysis, where chest wall displacement of 5-10 mm produces detectable Doppler shifts even when the subject is otherwise stationary.
Micro-Doppler Classification
Every moving target imposes a Doppler frequency shift on the reflected radar signal proportional to its radial velocity. But complex targets like humans produce multiple simultaneous Doppler shifts from different body parts: the torso moves at the walking velocity, while arms and legs swing at higher speeds, creating a distinctive frequency spread called the micro-Doppler signature. This signature is as unique to the type of motion as a fingerprint is to an individual.
Micro-Doppler classification uses spectrogram analysis, computing a time-frequency representation of the radar return that visualizes the characteristic patterns of walking, running, crawling, and animal movement. Convolutional neural networks trained on libraries of micro-Doppler spectrograms achieve classification accuracies above 92% in distinguishing human walking from animal movement, and above 85% in distinguishing between different human activities such as walking, running, and carrying loads.
Range-Doppler Processing
The fundamental signal processing chain in FMCW radar produces a two-dimensional Range-Doppler map. The first FFT across the fast-time samples of each chirp converts beat frequencies to range bins. The second FFT across the slow-time samples from successive chirps extracts Doppler velocity for each range bin. The resulting 2D map shows every target as a peak at its corresponding range and velocity coordinates.
Constant False Alarm Rate (CFAR) detection algorithms scan the Range-Doppler map to identify peaks that exceed the local noise floor by a statistically significant margin. Cell-averaging CFAR (CA-CFAR) estimates the noise level by averaging neighboring cells, while ordered-statistic CFAR (OS-CFAR) uses sorted neighbor values to provide more robust detection in environments with multiple closely spaced targets.
Integration in GoatArma™ Systems
GoatArma™ integrates a dual-band radar architecture into the CapraHelm™. A 77 GHz mmWave module provides high-resolution detection of nearby targets in the forward hemisphere, ideal for identifying individuals and tracking their movements in open environments. A separate UWB radar module operating at 3-5 GHz provides through-wall detection capability for building-entry operations, enabling the operator to detect occupants before crossing a threshold.
Both radar streams feed into the sensor fusion pipeline alongside LiDAR, thermal, and visual data. The radar's all-weather, through-obstacle capability fills critical gaps in the optical sensor coverage, ensuring that the operator maintains situational awareness even in degraded visual environments. The combined radar power consumption of approximately 4 watts is managed by the system's AI-based power management, which can selectively disable the TWR module during outdoor operations where its capability is less critical.