Human identification is crucial in various applications, including terrorist attack preventing, criminal seeking, defence and so on. Traditional human identification methods are usually based on vision, biological features, radio‐frequency identification cards and so on. In this study, the authors propose an identification method based on radar micro‐Doppler signatures using deep convolutional neural networks (DCNNs) for the first time, which can identify human in non‐contact, remote and no lighting status. They employ a K‐band Doppler radar to acquire the raw signals due to its stationary clutter rejection and movement detection ability as well as its short wavelength which can generate larger Doppler shift. Then short‐time Fourier transform is applied to the raw signals to characterise micro‐Doppler signatures. They adopt the DCNNs to deal with the spectrograms for human identification problem. The DCNNs can learn the necessary features and classification conditions from raw micro‐Doppler spectrograms without employing any explicit features. While the traditional supervised learning techniques relying on the extracted features require domain knowledge of each problem. It is shown that this method can achieve average accuracy ∼97.1% for 4 people, 90.9% for 6 people, 89.1% for 8 people, 85.6% for 10 people, 77.4% for 12 people, 72.6% for 16 people and 68.9% for 20 people.

Noncontact vital-sign telemonitoring by radar systems is applied more widely. When considering the implementation of Doppler radar for more precise extraction of cardiac physiological parameters to adapt to clinical diagnosis of heart diseases, the most open heart rate (HR) detection methods based on the periodic information are far from enough. Here, this article presents a new simple, effective, and high-accuracy settlement to maintain cardiac activity sensing comprehensively and systematically for the first time, for the estimation of beat-to-beat intervals (BBIs) and cardiac timings. We propose the decoding peak detection (DPD) method to address heartbeat peaks’ extraction problem by decoding the most likely sequence of states from the single-band frequency envelogram (FEnv) of the radar signal, thus realizing HR variability analysis. Furthermore, a hidden semi-Markov model (HSMM) is utilized to detect cardiac timings in bandpass filtered radar data, which is equivalent to positions of the R-peak and end-T-wave carried in electrocardiogram (ECG). Experimental results show that the proposed method achieves an average $F_{1}$ score of 93.19% ± 0.73% with the ECG as gold-standard, evaluated on a data set from six subjects. The accuracy of radar-based extracted BBIs is assessed by the mean relative error, which ranges from 0.51% to 1.06% by HSMM and 0.37% to 1.15% by DPD, respectively. Finally, a series of experiments are conducted to confirm the high agreement between radar signals and cardiac activities. Meanwhile, from the cardiac timings analysis by the proposed methods, we can find the advantage of radar-based detection at some specific timings compared with ECG-based, e.g., end-T-wave detection.

This paper presents different approximate designs for computing the FFT. The tradeoff between accuracy and performance is achieved by adjusting the word length in each computational stage. Two algorithms for word length modification under a specific error margin are proposed. The first algorithm targets an approximate FFT for an area-limited design compared to the conventional fixed design; the second algorithm targets performance so it achieves a higher operating frequency. Both of the proposed algorithms show that an efficient balance between hardware utilization and performance is possible at stage-level. The proposed approximate FFT designs are implemented on FPGA; experimental results show that hardware utilization using the first approximate algorithm are reduced by at least nearly 40%. The second algorithm increases performance of the designs by over 20%. Fine granularity design is also investigated, where the FPGA resources for a 256-point FFT computation can be further reduced by nearly 10% compared to a coarse design. Finally, the proposed approximate designs are applied to a feature extraction module in an isolated word recognition system; the numbers of LUTs and FFs for the Mel frequency cepstrum coefficients (MFCC) extraction module are decreased by up to 47.2% and 39.0%, respectively with a power reduction of up to 27.0% at a loss in accuracy of less than 2%.

Human identification based on radar signatures of individual heartbeats is crucial in various applications, including user authentication in mobile devices, identification of escaped criminals, etc. Usually, optical systems employed to recognize humans are sensitive to ambient light environments, while radar does not have such a drawback, since it has high penetration and all-weather capability. Meanwhile, since micro-Doppler characteristics from the heart of different people are distinct and not easy to fake, it can be used for identification. In this paper, we employed a deep convolutional neural network (DCNN) and conventional supervised learning methods to realize heartbeat-based identification. First, the heartbeat signals were acquired by a Doppler radar and processed by short-time Fourier transform. Then, predefined features were extracted for the conventional supervised learning algorithms, while time–frequency graphs were directly inputted to the DCNN since the network had its own feature extraction part. It is shown that the DCNN could achieve average accuracy of 98.5% for identifying four people, and higher than 80% when the number of people was less than ten. For conventional supervised learning algorithms when identifying four people, the accuracy of the support vector machine (SVM) was 88.75%, and the accuracy of SVM–Bayes was 91.25%, while naive Bayes had the lowest accuracy of 80.75%.

Deep learning (DL) is widely used in automatic target recognition (ATR) of synthetic aperture radar (SAR) images. Related researches show that DL models for SAR ATR are vulnerable to adversarial examples attack in the digital domain. However, how to generate adversarial examples in practical scenarios is critical and challenging. In this paper, we propose a systematic SAR perturbation generation algorithm (SAR-PeGA) for target recognition network. Firstly, assuming that some reflection phase tuning samples are located in the fixed area of SAR target, we adjust the phase characteristics of reflected signal with variable phase sequences. Secondly, we take the imperceptible perturbations from universal adversarial perturbations (UAP) as reference. Then, we construct the unconstrained minimum optimization model to find the specific phase sequences of tuning samples, and optimize the model with the adaptive moment estimation (Adam) optimizer. Finally, SAR adversarial examples can be flexibly generated through the proposed deceptive jamming model. Experimental results demonstrate that the proposed method can generate imperceptible jamming and effectively attack three classical recognition models.