Shelly Vishwakarma

In critical domains such as healthcare and law, accurately modelling the uncertainty of automatic computational models is essential. For instance, healthcare models must produce reliable estimates to guide human decision-making. However, modelling uncertainty remains challenging, particularly for models handling low-resource datasets and complex, domain-specific vocabulary. Most existing predictive models model point estimates rather than probability distributions, limiting our ability to quantify model uncertainty. This paper introduces a novel model, BAKER, designed to address these limitations. BAKER combines the strengths of Bayesian inference, known for its effectiveness in modelling uncertainty, and kernel methods, which excel at capturing complex data relationships. Incorporating kernel functions enhances model performance, particularly by reducing overfitting in data-limited scenarios. Our experimental analysis shows that BAKER significantly improves uncertainty reasoning compared to existing models.

Voice identification is being increasingly adopted in various domains, including security infrastructures, intelligent home systems, and personalised digital assistants. Notably, it harbours significant promise in transforming healthcare, especially in electronic health record detecting and speech impairment monitoring such as aphasia. Current strategies such as acoustic models based on deep learning, voice bio‐metrics, and spectrogram analysis, have been identified with several drawbacks including vulnerability to altered voices, susceptibility to ambient noise, and the necessity for significant computational power. In response to these issues, the authors introduce a ground‐breaking method of voice identification using Ultra‐Wideband (UWB) technology. This method capitalises on the micro‐Doppler shifts associated with movements of the laryngeal prominence. The distinctive nature of these bio‐metric traits related to speech production provides superior resistance against common pitfalls of voice identification. The proposed model leverages the high‐resolution characteristics of UWB to register tiny variations in laryngeal movements produced during speech, thus forming a distinct voice profile for each speaker. Through rigorous testing, the proposed system demonstrated significant progress in voice identification, achieving close to 90% accuracy in controlled experimental settings. This breakthrough indicates that UWB‐enabled voice identification could have a profound effect on medical applications, providing potential improvements in diagnosing, monitoring, possibly treating speech disorders, and thereby shaping a future of enhanced and secured healthcare services.

Motion tracking systems based on optical sensors typically suffer from poor lighting conditions, occlusion, limited coverage, and may raise privacy concerns. More recently, radiofrequency (RF) based approaches using commercial WiFi devices have emerged which offer low-cost ubiquitous sensing whilst preservin privacy. However, RF sensing systems typically output range-Doppler maps, time-frequency spectrograms, cross-range plots etc which cannot represent human motion intuitively and usually requires further processing. In this study, we propose MDPose, a novel framework for human skeletal motion reconstruction base on WiFi micro-Doppler signatures. MDPose provides an effective solution to represent human activity by reconstructing a skeleton model with 17 key points, which can assist with the interpretation of conventional RF sensing outputs in a more understandable way. Specifically, MDPose is implemented over three sequential stage to address a series of challenges: First, a denoising algorithm is employed to remove any unwanted noise that may affect feature extraction and enhance weak Doppler measurements. Secondly, a convolutional neural network (CNN)-recurrent neural network (RNN) architecture is applied to learn temporal spatial dependenc from clean micro-Doppler signatures and restore velocity information to key points under the supervision of the motion capture (Mocap) system. Finally, a pose optimisation mechanism based on learning optimisation vectors is employed to estimate the initial state of the skeleton and to limit additional errors. We hav conducted a comprehensive set of tests in a variety of environments using numerous subjects with a single receiver radar system to demonstrate the performance of MDPose, and report 29.4mm mean absolute error over all key points positions on several common daily activities, which has performance comparable to that of state-ofthe- art RF-based pose estimation systems.

This work introduces an attention mechanism that can be integrated into any standard convolution neural network to improve model sensitivity and prediction accuracy with minimal computational overhead. The attention mechanism is introduced in a lightweight network – Alexnet and its classification performance for human micro‐Doppler signatures is evaluated. The Alexnet model trained with an attention module can implicitly highlight the salient regions in the radar signatures while suppressing the irrelevant background regions and consistently improving network predictions. Network visualizations are provided through class activation mapping, providing better insights into how the predictions are made. The visualizations demonstrate how the attention mechanism focusses on the region of interest in the radar signatures.

This paper presents a comprehensive dataset intended to evaluate passive Human Activity Recognition (HAR) and localization techniques with measurements obtained from synchronized Radio-Frequency (RF) devices and vision-based sensors. The dataset consists of RF data including Channel State Information (CSI) extracted from a WiFi Network Interface Card (NIC), Passive WiFi Radar (PWR) built upon a Software Defined Radio (SDR) platform, and Ultra-Wideband (UWB) signals acquired via commercial off-the-shelf hardware. It also consists of vision/Infra-red based data acquired from Kinect sensors. Approximately 8 hours of annotated measurements are provided, which are collected across two rooms from 6 participants performing 6 daily activities. This dataset can be exploited to advance WiFi and vision-based HAR, for example, using pattern recognition, skeletal representation, deep learning algorithms or other novel approaches to accurately recognize human activities. Furthermore, it can potentially be used to passively track a human in an indoor environment. Such datasets are key tools required for the development of new algorithms and methods in the context of smart homes, elderly care, and surveillance applications.

Software defined radar (SDRadar) systems have become an important area for future radar development and are based on similar concepts to Software defined radio (SDR). Most of the processing like filtering, frequency conversion and signal generation are implemented in software. Currently, radar systems tend to have complex signal processing and operate at wider bandwidth, which means that limits on the available computational power must be considered when designing a SDRadar system. This paper presents a feasible solution to this potential limitation by accelerating the signal processing using a GPU to enable the development of a high speed SDRadar system. The developed system overcomes the limitation on the processing speed by CPU‐only, and has been tested on three different SDR devices. Results show that, with GPU accelerator, the processing rate can achieve up to 80 MHz compared to 20 MHz with the CPU‐only. The high speed processing makes it possible to run in real‐time and process full bandwidth across the WiFi signal acquired by multiple channels. The gains made through porting the processing to the GPU moves the technology towards real‐world application in various scenarios ranging from healthcare to IoT, and other applications that required significant computational processing.

Motion tracking systems based on optical sensors typically often suffer from issues, such as poor lighting conditions, occlusion, limited coverage, and may raise privacy concerns. More recently, radio frequency (RF)-based approaches using commercial WiFi devices have emerged which offer low-cost ubiquitous sensing whilst preserving privacy. However, the output of an RF sensing system, such as Range-Doppler spectrograms, cannot represent human motion intuitively and usually requires further processing. In this study, MDPose, a novel framework for human skeletal motion reconstruction based on WiFi micro-Doppler signatures, is proposed. It provides an effective solution to track human activities by reconstructing a skeleton model with 17 key points, which can assist with the interpretation of conventional RF sensing outputs in a more understandable way. Specifically, MDPose has various incremental stages to gradually address a series of challenges: First, a denoising algorithm is implemented to remove any unwanted noise that may affect the feature extraction and enhance weak Doppler signatures. Secondly, the convolutional neural network (CNN)-recurrent neural network (RNN) architecture is applied to learn temporal-spatial dependency from clean micro-Doppler signatures and restore key points' velocity information. Finally, a pose optimising mechanism is employed to estimate the initial state of the skeleton and to limit the increase of error. We have conducted comprehensive tests in a variety of environments using numerous subjects with a single receiver radar system to demonstrate the performance of MDPose, and report 29.4mm mean absolute error over all key points positions, which outperforms state-of-the-art RF-based pose estimation systems.

Radio-frequency-based noncooperative monitoring of humans has numerous applications ranging from law enforcement to ubiquitous sensing applications such as ambient assisted living and biomedical applications for nonintrusively monitoring patients. Large training datasets, almost unlimited memory capacity, and ever-increasing processing speeds of computers could drive forward the data-driven deep-learning-focused research in the abovementioned applications. However, generating and labeling large volumes of high-quality, diverse radar datasets is an onerous task. Furthermore, unlike the fields of vision and image processing, the radar community has limited access to databases that contain large volumes of experimental data. Therefore, in this article, we present an open-source motion capture data-driven simulation tool, SimHumalator, that can generate large volumes of human micro-Doppler radar data in passive WiFi scenarios. The simulator integrates IEEE 802.11 WiFi Standards (IEEE 802.11 g, n, and ad) compliant transmissions with the human animation data to generate the micro-Doppler features that incorporate the diversity of human motion characteristics and the sensor parameters. The simulated signatures have been validated with experimental data gathered using an in-house-built hardware prototype. This article describes simulation methodology in detail and provides case studies on the feasibility of using simulated micro-Doppler spectrograms for data augmentation tasks.

IoT ecosystems consist of a range of smart devices that generated a plethora of Radio Frequency (RF) transmissions. This provides an attractive opportunity to exploit already-existing signals for various sensing applications such as e-Healthcare, security and smart home. In this paper, we present Passive IoT Radar (PIoTR), a system that passively uses RF transmissions from IoT devices for human monitoring. PIoTR is designed based on passive radar technology, with a generic architecture to utilize various signal sources including the WiFi signal and wireless energy at the Industrial, Scientific and Medical (ISM) band. PIoTR calculates the phase shifts caused by human motions and generates Doppler spectrogram as the representative. To verify the proposed concepts and test in a more realistic environment, we evaluate PIoTR with four commercial IoT devices for home use. Depending on the effective signal and power strength, PIoTR performs two modes: coarse sensing and fine-grained sensing. Experimental results show that PIoTR can achieve an average of 91% in occupancy detection (coarse sensing) and 91.3% in activity recognition (fine-grained sensing).

Micro-Doppler signatures contain considerable information about target dynamics. However, the radar sensing systems are easily affected by noisy surroundings, resulting in uninterpretable motion patterns on the micro-Doppler spectrogram ( $\mu $ -DS). Meanwhile, radar returns often suffer from multipath, clutter, and interference. These issues lead to difficulty in, for example, motion feature extraction and activity classification using micro-Doppler signatures. In this article, we propose a latent feature-wise mapping strategy, called feature mapping network (FMNet), to transform measured spectrograms so that they more closely resemble the output from a simulation under the same conditions. Based on measured spectrogram and the matched simulated data, our framework contains three parts: an encoder which is used to extract latent representations/features, a decoder outputs reconstructed spectrogram according to the latent features, and a discriminator minimizes the distance of latent features of measured and simulated data. We demonstrate the FMNet with six activities data and two experimental scenarios, and final results show strong enhanced patterns and can keep actual motion information to the greatest extent. On the other hand, we also propose a novel idea which trains a classifier with only simulated data and predicts new measured samples after cleaning them up with the FMNet. From final classification results, we can see significant improvements.