Radar systems are increasingly being employed in healthcare applications for human activity recognition due to their advantages in terms of privacy, contactless sensing, and insensitivity to lighting conditions. The proposed classification algorithms are however often complex, focusing on a single domain of radar, and requiring significant computational resources that prevent their deployment in embedded platforms which often have limited memory and computational resources. To address this issue, we present an adaptive magnitude thresholding approach for highlighting the region of interest in the multi-domain micro-Doppler signatures. The region of interest is beneficial to extract salient features, meanwhile it ensures the simplicity of calculations with less computational cost. The results for the proposed approach show an accuracy of up to 93.1% for six activities, outperforming state-of-the-art deep learning methods on the same dataset with an over tenfold reduction in both training time and memory footprint, and a twofold reduction in inference time compared to a series of deep learning implementations. These results can help bridge the gap toward embedded platform deployment.

@en

Radar micro-Doppler signatures have been proposed for human monitoring and activity classification for surveillance and outdoor security, as well as for ambient assisted living in healthcare-related applications. A known issue is the performance reduction when the target is moving tangentially to the line of sight of the radar. Multiple techniques have been proposed to address this, such as multistatic radar and to some extent, interferometric (IF) radar. A simulator is presented to generate synthetic data representative of eight radar systems (monostatic, circular multistatic and in-line multistatic [IM] and IF) to quantify classification performances as a function of aspect angles and deployment geometries. This simulator allows an unbiased performance evaluation of different radar systems. Six human activities are considered with signatures originating from motion-captured data of 14 different subjects. The classification performances are analysed as a function of aspect angles ranging from 0° to 90° per activity and overall. It demonstrates that IF configurations are more robust than IM configurations. However, IM performs better at angles below 55° before IF configurations take over.

@en

In personnel recognition based on radar, significant research exists on statistical features extracted from the micro-Doppler signatures, whereas research considering other domains and information such as phase is less developed. This paper presents the use of deep learning methods to integrate both phase and magnitude features from range profiles and spectrogram. The temporal features of both domains are separately extracted using a stack of Long Short Term Memory (LSTM) layers. Then, the extracted features are aggregated in the corresponding domains and pass through a series of dense layers with SoftMax classifier. Finally, the information from the two domains is fused with a soft fusion approach to improve the performance further. Preliminary results show that the proposed network with soft fusion can achieve 85.5% accuracy in personnel recognition with six subjects@en

As the number of older adults increases worldwide, new paradigms for indoor activity monitoring are required to keep people living at home independently longer. Radar-based human activity recognition has been identified as a sensing modality of choice because it is privacy-preserving and does not require end-users compliance or manipulation. In this paper, we explore the robustness of machine learning algorithms for human activity recognition using six different activities from the University of Glasgow dataset recorded with an FMCW radar. The raw radar data is pre-processed and represented using four different domains, namely, range-time, range-Doppler amplitude and phase diagrams, and Cadence Velocity Diagram. From those, salient features can be extracted and classified using Support Vector Machine, Stacked AutoEncoder, and Convolutional Neural Networks. The fusion of handcrafted features and features from CNN is applied to get the best scheme of classification with over 96% accuracy.@en

Radar-based classification of human activities and gait have attracted significant attention with a large number of approaches proposed in terms of features and classification algorithms. A common approach in activity classification attempts to find the algorithm (features plus classifier) that can deal with multiple activities analysed in one study such as walking, sitting, drinking and crawling. However, using the same set of features for multiple activities can be suboptimal per activity and not take into account the diversity of kinematic movements that could be captured by diverse features. In this paper, we propose a hierarchical classification approach that uses a large variety of features including but not limited to energy features like entropy and energy curve, physical features like centroid and bandwidth, image-based features like skewness extracted from multiple radar data domains. Feature selection is used at each step of the hierarchical model to select the best set of features to discriminate the target activity from the others, showing improvements with respect to the more conventional approach of using a multiclass model. The proposed approach is validated on a large dataset with 1078 recorded samples of varying length from 5 s to 10 s of experimental data, yielding 95.4% accuracy to classify six activities. The approach is also validated on a personnel recognition task to identify individual subjects from their walking gait, yielding 83.7% accuracy for ten subjects and 68.2% for a significantly larger group of subjects, i.e., 60 people.

@en

Nowadays, health monitoring issues are increasing as the worldwide population is aging. In this paper, the radar modality is used to classify with radar signature automatically. The classic approach is to extract features from micro-Doppler signatures for classification. This data representation domain has its limitations for activities presenting similar accelerations like a frontal fall and picking up an object from the floor that lead to wrongly labeled activities. In this work, we propose to combine multiple radar data domains with deep learning. Features are extracted from four domains, namely, Range-Time, Range-Doppler, Doppler-Time, and Cadence Velocity Diagram. The extracted features are set as the input of a Convolutional Neural Network, yielding 91% accuracy with 10-fold cross-validation based on the University of Glasgow “Radar signatures of human activities” open dataset.@en