Dr. Nir Regev
I build statistical signal processing and AI algorithms for radar, lidar, ultrasound and other sensors, from the noise model to the detector, tracker and classifier that run on the device. I also bring the same estimation and control tools to multidisciplinary work in medicine and biology.
- Chief Scientist and Co-Founder, AlephZero.ai
- Adjunct Professor, Electrical and Computer Engineering, Cal Poly Pomona
- Founder, Regev Radar and AI Academy
- Education
- Ph.D. in ECE, Ben-Gurion University
- Field
- Radar, lidar and statistical signal processing
- Experience
- 30 years
About
I have spent 30 years building signal processing and machine learning systems for radar, lidar, ultrasound and embedded sensing. My Ph.D. in Electrical and Computer Engineering at Ben-Gurion University focused on remote sensing of human vital signs with radar, and that thread still runs through my work.
At AlephZero.ai I lead the algorithm work behind mmWave radar ambient intelligence: contactless vital signs, sleep sensing, fall detection and target classification. I teach probability and control at Cal Poly Pomona, and I run the Regev Radar and AI Academy, a community where engineers and students learn radar signal processing and AI together.
I am a Senior Member of the IEEE and speak often at IEEE and AI conferences.
I also consult for engineering teams on radar and AI signal chains, and serve as a technical expert in patent and IP matters.
Current research
Contactless vital signs and sleep
Breathing and heart rate from mmWave and UWB radar, joint presence detection and rate estimation, and sleep monitoring without wearables.
FMCW processing and estimation
Range and Doppler processing, windowing and sidelobe control, peak interpolation, and estimator performance measured against the Cramér-Rao bound.
Detection in real clutter
CFAR detectors for non-homogeneous backgrounds and the tradeoffs between false-alarm control and detection loss.
Tracking and learned filters
Kalman and IMM trackers, and fair comparisons between learned filters such as KalmanNet and well-tuned model-based baselines.
Target classification
Micro-Doppler signatures and track-level features for classifying people, drones and vehicles with compact neural networks.
Higher-order statistics
Higher-order ambiguity functions and cumulants for estimating polynomial-phase signals in noise.
Biological signal propagation
Modeling how immune signals such as IL-2, IL-22 and alpha-1 antitrypsin (AAT) propagate and drive tissue injury and repair, with nonlinear state-space models and the estimation tools of radar tracking.
Closed-loop physiology
Control-theoretic analysis of insulin dosing, including why fixed-gain correction insulin fails when the body's own insulin secretion is intact.
Publications
-
2026
Why fixed-gain correction insulin fails when endogenous insulin secretion is intact: a closed-loop analysis Preprintdoi:10.21203/rs.3.rs-11273023/v1
-
2023
Analog, in-memory compute architectures for artificial intelligence PreprintarXiv:2302.06417
-
2021
Radar-based, simultaneous human presence detection and breathing rate estimationdoi:10.3390/s21103529
-
2020
Multi-modal, remote breathing monitordoi:10.3390/s20041229
-
2020
A simple, remote, ultrasonic based personal emergency response systemdoi:10.1109/WMCS49442.2020.9172404
-
2019
Remote sensing of vital signs using an ultra-wide-band radardoi:10.1080/2150704X.2019.1573335
-
2019
On direction of arrival estimation with 1-bit quantizerdoi:10.1109/RADAR.2019.8835785
-
2017
Classification of single and multi propelled miniature drones using multilayer perceptron artificial neural networkdoi:10.1049/cp.2017.0378
-
2017
On optimal receiver for nonlinearly distorted single carrier signaldoi:10.1109/PIMRC.2017.8292168
-
2017
A simple, remote, video based breathing monitordoi:10.1109/EMBC.2017.8037191
-
2015
Maximum likelihood detection of nonlinearly distorted OFDM signaldoi:10.1109/GLOCOM.2014.7417009
Patents
Granted
-
US 10,856,746
System and method for estimating vital signs
Published applications
-
US 2024/0159881 A1
FMCW lidar signal disambiguation sampling and processing
-
US 2018/0064373 A1
Detecting falls and near falls by ultrasound
-
WO 2017/195196 A1
Remote estimation of vital signs
Pending
-
U.S. Prov. 64/160,853
Patient-specific modeling of glucose regulation with intact endogenous insulin secretion, and evaluating, reconstructing and replacing insulin correction rules
-
U.S. Prov. 64/161,791
Scheduling intermittent infusions that carry a fixed daily substance load to minimize peak delivery and peak physiological response
Earlier provisional applications
- U.S. Prov. 62/333,298Remotely estimating vital signs of a subject
- U.S. Prov. 62/374,936Recognizing gestures by ultrasound
- U.S. Prov. 62/383,443Detecting falls and near falls by ultrasound
- U.S. Prov. 62/383,445Monitoring baby breathing rate
- U.S. Prov. 62/625,578Method of remotely monitoring respiration
Teaching and talks
Cal Poly Pomona
- ECE 3715: Probability, Statistics, and Random Processes
- ECE 3709: Control Systems Engineering
- Advisor to two senior design teams and to four student teams at the CPP AI Conference
Talks and tutorials
- Half-day tutorial, IEEE Radar Conference (RadarConf) 2026
- AI in Radar Signal Processing, Embedded Online Conference
- The Micro-Doppler Effect in Radar, Embedded Online Conference
Regev Radar and AI Academy
The academy is a community of engineers and students working on radar, lidar and other sensing systems, learning together. Members ask questions, bring their own problems and work through them with each other and with me. Inside it are my courses, hands-on labs and live lectures, all built the way I practice: start from the signal model and the noise, derive the estimator or detector, then build it and measure how well it works. It is for people who want the reasoning behind the algorithms, not only the recipes.
Courses
Practical FMCW Radar Signal Processing
The FMCW chain from chirp to range-Doppler map, and the choices that set resolution, ambiguity and sensitivity.
AI in Radar Signal Processing
Where learned models beat classical processing in radar, where they do not, and how to tell the difference.
Peak Location Estimation in Radar and Lidar
Getting sub-bin accuracy out of a spectral peak, and how close an estimator gets to the theoretical limit.
HAF Transform in Radar
Higher-order ambiguity functions for estimating polynomial-phase signals such as chirps and maneuvering targets.
Understanding the ROC
Detection probability, false alarms and thresholds, read directly from the receiver operating characteristic.
And many more…
New courses, labs and live sessions keep coming. See them all in the Academy →
Books
- Understanding Probability with Python
- Understanding Probability with MATLAB
- Soft, Hands-On Introduction to Artificial Intelligence with Python
Learn radar and AI with me
A community of engineers and students, with courses on FMCW processing, estimation, detection and AI for radar, taught by someone who ships these systems.
Contact
For consulting, expert-witness work, talks or collaboration, write to contact@drnirregev.com.
