Abstract-This paper reviews recent works in the literature on the use of systems based on Radar and RGB-Depth sensors for fall detection, and discusses outstanding research challenges and trends related to this research field. Systems to detect reliably fall events and promptly alert carers and first responders have gained significant interest in the past few years in order to address the societal issue of an increasing number of elderly people living alone, with the associated risk of them falling and the consequences in terms of health treatments, reduced well-being, and costs. The interest in radar and RGB-D sensors is related to their capability to enable contactless and non-intrusive monitoring, which is an advantage for practical deployment and users' acceptance and compliance, compared with other sensor technologies such as video-cameras, or wearables. Furthermore, the possibility of combining and fusing information from heterogeneous types of sensors is expected to improve the overall performance of practical fall detection systems. Researchers from different fields can benefit from multidisciplinary knowledge and awareness of the latest developments in radar and RGB-D sensors that this paper is discussing.

In this work, we present results for classification of different classes of targets (car, single and multiple people, bicycle) using automotive radar data and different neural networks. A fast implementation of radar algorithms for detection, tracking, and micro-Doppler extraction is proposed in conjunction with the automotive radar transceiver TEF810X and microcontroller unit SR32R274 manufactured by NXP Semiconductors. Three different types of neural networks are considered, namely a classic convolutional network, a residual network, and a combination of convolutional and recurrent network, for different classification problems across the 4 classes of targets recorded. Considerable accuracy (close to 100% in some cases) and low latency of the radar pre-processing prior to classification (approximately 0.55s to produce a 0.5s long spectrogram) are demonstrated in this paper, and possible shortcomings and outstanding issues are discussed.

Growing life expectancy and increasing incidence of multiple chronic health conditions are significant societal challenges. Different technologies have been proposed to address these issues, to detect critical events such as stroke or falls, and to monitor automatically human activities for health condition inference and anomalies detection. This paper aims to investigate two types of sensing technologies proposed for assisted living: wearable and radar sensors. First, different feature selection methods are validated and compared in terms of accuracy and computational loads. Then, information fusion is applied to enhance activity classification accuracy combining the two sensors. Improvements in classification accuracy of approximately 12% using feature level fusion is achieved with both Support Vector Machine and K Nearest Neighbor classifiers. Decision-level fusion schemes are also investigated, yielding classification accuracy in the order of 97-98%.

This review explores radar based techniques currently utilised in literature to monitor small UAVs or drones; several challenges have arisen due to their rapid emergence and commercialisation within the mass market. The potential security threats posed by these systems are collectively presented and the legal issues surrounding their successful integration is briefly outlined. Key difficulties involved in the identification and hence tracking of these 'radar elusive' systems are discussed, along with how research efforts relating to drone detection, classification and RCS characterisation are being directed in order to address this emerging challenge. Such methods are thoroughly analysed and critiqued; finally, an overall picture of the field in its current state is painted, alongside scope for future work over a broad spectrum.

No abstract

This article presents radar signal processing for sensing in the context of assisted living. This is covered through 3 example applications: human activity recognition for activities of daily living, respiratory disorder and Sleep Stages classification. The common challenge of classification is discussed within a framework of measurements/pre-processing, feature extraction, and classification algorithms for supervised learning. Then, the specific challenges of the 3 applications from a signal processing standpoint are detailed in their specific data processing and ad-hoc classification strategies, focusing on recent trends in the field of activity recognition (multidomain, multi-modal and fusion) and healthcare applications based on vital signs (super-resolution techniques) and commenting on outstanding challenges. To conclude, this paper explores the challenge of the real-time implementation of signal processing/classification algorithms.

This paper presents the use of micro-Doppler signatures collected by a multistatic radar to detect and discriminate between micro-drones hovering and flying while carrying different payloads, which may be an indication of unusual or potentially hostile activities. Different features have been extracted and tested, namely features related to the Radar Cross Section of the micro-drones, as well as the Singular Value Decomposition (SVD) and centroid of the micro-Doppler signatures. In particular, the added benefit of using multistatic information in comparison with conventional radar is quantified. Classification performance when identifying the weight of the payload that the drone was carrying while hovering was found to be consistently above 96% using the centroid-based features and multistatic information. For the non-hovering scenarios classification results with accuracy above 95% were also demonstrated in preliminary tests in discriminating between three different payload weights.

Classification of different human activities using multistatic micro-Doppler data and features is considered in this paper, focusing on the distinction between unarmed and potentially armed personnel. A database of real radar data with more than 550 recordings from 7 different human subjects has been collected in a series of experiments in the field with a multistatic radar system. Four key features were extracted from the micro-Doppler signature after Short Time Fourier Transform analysis. The resulting feature vectors were then used as individual, pairs, triplets, and all together before inputting to different types of classifiers based on the discriminant analysis method. The performance of different classifiers and different feature combinations is discussed aiming at identifying the most appropriate features for the unarmed vs armed personnel classification, as well as the benefit of combining multistatic data rather than using monostatic data only. 1-IntroductionThis paper presents the analysis of radar micro-Doppler signatures from a multistatic radar system. The data was generated using NetRAD [1] which is a three-node multistatic radar system that has been developed over the last decade at University College London. The system has been adapted since 2007 to a higher power wireless configuration to increase the flexibility of the measurement possibilities [2], and has provided interesting and novel results in the field of bistatic sea clutter characterisation and analysis [3].Radar Micro-Doppler is the phenomenon of the observed micro-motions on top of the bulk main Doppler component of a target's motion. It has been the subject of research over a number of years focusing on the additional information that can be extracted from this signal.Such information can then be exploited in a variety of applications for security, law enforcement, urban warfare, and search and rescue, where the detection, tracking, and classification of many human targets moving in a cluttered environment is of paramount importance.