In this article, a new microwave-radar-based technique for short-range detection and classification of multiple human and vehicle targets crossing a monitored area is proposed. This approach, which can find applications in both security and infrastructure surveillance, relies upon the processing of the scattered-field data acquired by low-cost off-the-shelf components, i.e., a 24 GHz frequency-modulated continuous wave (FMCW) radar module and a Raspberry Pi mini-PC. The developed method is based on an ad hoc processing chain to accomplish the automatic target recognition (ATR) task, which consists of blocks performing clutter and leakage removal with an infinite impulse response (IIR) filter, clustering with a density-based spatial clustering of applications with noise (DBSCAN) approach, tracking using a Benedict-Bordner $\alpha $ - $\beta $ filter, features extraction, and finally classification of targets by means of a $k$ -nearest neighbor ( $k$ -NN) algorithm. The approach is validated in real experimental scenarios, showing its capabilities in correctly detecting multiple targets belonging to different classes (i.e., pedestrians, cars, motorcycles, and trucks).

The WIVERN (WInd VElocity Radar Nephoscope) mission concept, a candidate for ESA’s Earth Explorer 11 program, aims at globally observing vertical profiles of reflectivity and line-of-sight winds in cloudy and precipitating regions. WIVERN employs a 94 GHz dual-polarisation Doppler radar with conical scanning to address the limited coherence duration between radar transmitted from low-Earth satellites with small antennas. This system transmits closely spaced pairs of horizontally and vertically polarised pulses, which are better correlated than pulses of the same polarization separated by longer intervals. The Polarisation Diversity Pulse Pair (PDPP) technique is then used to estimate radar observables like reflectivities, differential reflectivities, Doppler velocities, and differential phase. The paper introduces an efficient method for generating H- and V-I&Q time series from the covariance matrix of the autocorrelation function. This method treats the signal as a non-stationary stochastic process, making it suitable for the PDPP pulse sequence from a rapidly rotating antenna and more computationally efficient than inverse fast Fourier transform techniques. It also accounts for interfering cross-polar signals and decorrelation from the scanning antenna. This method is included in the mission’s end-to-end simulator, which processes data from raw I&Q to Level 1 estimates of polarimetric variables. For scientific applications, averaging at least 5 km (40 polarisation diversity pairs) is necessary to reduce noise in polarimetric variables and Doppler velocities. Under optimal conditions, uncertainties at 5 km integration are 0.7 dB for reflectivities, 0.3 dB for Z_DR, 0.4 m/s for Doppler velocities, and 1.9° for Φ_DP.

In this article, we propose a radar-based human action recognition system, capable of recognizing actions in real time. Range-Doppler maps extracted from a low-cost frequency-modulated continuous wave (FMCW) radar are fed into a deep neural network. The system is deployed on an edge device. The results show that the system can recognize five human actions with an accuracy of 93.2% and an inference time of 2.95 s. Raising an alarm when a harmful action happens is a crucial feature in an indoor safety application. Thus, the performance during the binary classification, i.e., fall vs nonfall actions, is also assessed, achieving an accuracy of 96.8% with a false-negative rate of 4%. To find the best tradeoff between accuracy and computational cost, the energy precision ratio of the system deployed on the edge is measured. The system achieves a 1.04 energy precision ratio value, where an ideal ratio would be close to zero.

A prototype of a radar system for classification of multiple targets passing through a road gate is presented in this paper. It allows to identify different types of targets, i.e., pedestrians, motorcycles, cars, and trucks. The developed system is based on a low-cost 24 GHz off-the-shelf FMCW radar, combined with an embedded Raspberry Pi PC for data acquisition and transmission to a remote processing PC, which take care of detection and classification. The processing chain relies upon a tracking algorithm to follow the targets during traversal, combined with a classification scheme based on support vector machines. The approach has been validated with experimental data acquired in different scenarios, showing good identification capabilities.

The WIVERN (Wind VElocity Radar Nephoscope) mission, one of the four ESA Earth Explorer 11 candidate missions, aims at globally observing, for the first time, simultaneously vertical profiles of reflectivities and line of sight winds in cloudy and precipitating regions. WIVERN adopts a dual-polarization Doppler radar in order to overcome the short decorrelation time between successive radar pulses transmitted from low Earth-orbiting satellites with finite beamwidth antennas. WIVERN transmits a single polarization state at a time (H or V), receives in both polarization states, and uses the Polarization Diversity Pulse Pair (PDPP) technique to estimate the Doppler velocity. The weaker cross-polar signals can sometimes interfere with the co-polar ones, causing ghost signals in the measurements that hinder the system’s overall performance. Additionally, with the envisaged radar trigger mode, parameters such as Linear Depolarization Ratio (LDR) and Differential Reflectivity (Z _DR ) cannot be directly measured because of the nearly simultaneous transmission of H and V pulses. To overcome these challenges, this article presents a novel technique based on the Optimal Estimation (OE) algorithm for retrieving LDR, Z _DR , and co-polar reflectivity for radars operated in PDPP mode. The performance of the proposed method is evaluated using a realistic climatology of profiles simulated from CloudSat data. Results demonstrate that co-polar reflectivity can be accurately retrieved in regions with a good signal-to-noise ratio and in the absence of simultaneous cross-talk interference in both channels (which occurs very rarely). The LDR retrieval on the other hand is typically driven by the a-priori with a substantial impact of measurements only for the surface returns. The impact of cross-talk is also assessed on the reduction of precise Doppler measurements. Findings confirm that a selection of the separation between the two polarization diversity pulses ( T_HV ) of 20 μs achieves a good balance between the large errors originated by the strong dependence on the Doppler phase noise at small T_HV s and those caused by the drop in correlation and unambiguous Nyquist velocity at large T_HV .

In this paper, we analyze the radar echo signals of a 24 GHz FMCW radar obtained from two different groups of moving targets (humans and vehicles). Range and Doppler information are obtained with a 2D FFT analysis. The aim of this paper is to describe the experimental setup, the adopted approach for the extraction of the Doppler-based features for human-vehicle classification, and the results obtained so far. In particular, the differences between pedestrians and vehicles in the Doppler domain will be reported and discussed.