Publications
2023
Vishwakarma, Shelly; Li, Wenda; Tang, Chong; Woodbridge, Karl; Adve, Raviraj; Chetty, Kevin
Attention-enhanced Alexnet for improved radar micro-Doppler signature classification Journal Article
In: IET Radar, Sonar & Navigation, vol. 17, no. 4, pp. 652–664, 2023.
Abstract | Links | BibTeX | Tags:
@article{vishwakarma2023attention,
title = {Attention-enhanced Alexnet for improved radar micro-Doppler signature classification},
author = {Shelly Vishwakarma and Wenda Li and Chong Tang and Karl Woodbridge and Raviraj Adve and Kevin Chetty},
url = {https://doi.org/10.1049/rsn2.12369},
year = {2023},
date = {2023-01-01},
urldate = {2023-01-01},
journal = {IET Radar, Sonar & Navigation},
volume = {17},
number = {4},
pages = {652–664},
publisher = {Wiley Online Library},
abstract = {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.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
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.
Tang, Chong; Li, Wenda; Vishwakarma, Shelly; Shi, Fangzhan; Julier, Simon; Chetty, Kevin
MDPose: Human Skeletal Motion Reconstruction Using WiFi Micro-Doppler Signatures Journal Article
In: IEEE Transactions on Aerospace and Electronic Systems, pp. 1-12, 2023.
Abstract | Links | BibTeX | Tags:
@article{10068751,
title = {MDPose: Human Skeletal Motion Reconstruction Using WiFi Micro-Doppler Signatures},
author = {Chong Tang and Wenda Li and Shelly Vishwakarma and Fangzhan Shi and Simon Julier and Kevin Chetty},
doi = {10.1109/TAES.2023.3256973},
year = {2023},
date = {2023-01-01},
urldate = {2023-01-01},
journal = {IEEE Transactions on Aerospace and Electronic Systems},
pages = {1-12},
abstract = {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.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
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.
2022
Vishwakarma, Shelly; Li, Wenda; Adve, Raviraj; Chetty, Kevin
Learning salient features in radar micro-Doppler signatures using Attention Enhanced Alexnet Proceedings Article
In: International Conference on Radar Systems (RADAR 2022), pp. 190-195, 2022.
Abstract | Links | BibTeX | Tags:
@inproceedings{10038864,
title = {Learning salient features in radar micro-Doppler signatures using Attention Enhanced Alexnet},
author = {Shelly Vishwakarma and Wenda Li and Raviraj Adve and Kevin Chetty},
url = {https://ieeexplore.ieee.org/abstract/document/10038864},
doi = {10.1049/icp.2022.2314},
year = {2022},
date = {2022-01-01},
urldate = {2022-01-01},
booktitle = {International Conference on Radar Systems (RADAR 2022)},
volume = {2022},
pages = {190-195},
abstract = {This work introduces an attention mechanism that can be integrated into any standard convolution neural network (CNN) to improve model sensitivity and prediction accuracy with minimal computational overhead. We introduce the attention mechanism in a lightweight network-Alexnet and evaluate its classification performance for human micro-Doppler signatures. We show that the Alexnet model trained with an attention module can implicitly learn to highlight the salient regions in the radar signatures whilst suppressing the irrelevant background regions and consistently improve the network predictions by more than 4% in most cases. We further provide network visualizations through class activation mapping, providing better insights into how the predictions are made.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
This work introduces an attention mechanism that can be integrated into any standard convolution neural network (CNN) to improve model sensitivity and prediction accuracy with minimal computational overhead. We introduce the attention mechanism in a lightweight network-Alexnet and evaluate its classification performance for human micro-Doppler signatures. We show that the Alexnet model trained with an attention module can implicitly learn to highlight the salient regions in the radar signatures whilst suppressing the irrelevant background regions and consistently improve the network predictions by more than 4% in most cases. We further provide network visualizations through class activation mapping, providing better insights into how the predictions are made.
Vishwakarma, Shelly; Chetty, Kevin; others,
Realistic Micro-Doppler Database Generation Through Neural Style Transfer Framework Journal Article
In: 2022.
Abstract | Links | BibTeX | Tags:
@article{vishwakarma2022realistic,
title = {Realistic Micro-Doppler Database Generation Through Neural Style Transfer Framework},
author = {Shelly Vishwakarma and Kevin Chetty and others},
url = {https://discovery.ucl.ac.uk/id/eprint/10153700/},
year = {2022},
date = {2022-01-01},
urldate = {2022-01-01},
abstract = {Real-time monitoring of humans can assist professionals in providing healthy living enabling technologies to ensure the health, safety, and well-being of people of all age groups. To enhance the human activity recognition performance, we propose a style-transfer neural framework to generate realistic synthetic micro-Doppler signature dataset. The proposed network extracts environmental effects such as noise, multipath, and occlusions effects directly from the measurement data and transfers these features to the clean simulated signatures generated through our simulator called SimHumaLator. This results in more realistic-looking signatures qualitatively and quantitatively. We use these enhanced signatures to augment our measurement data and observe an improvement in the classification performance by 5% compared to no augmentation case.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Real-time monitoring of humans can assist professionals in providing healthy living enabling technologies to ensure the health, safety, and well-being of people of all age groups. To enhance the human activity recognition performance, we propose a style-transfer neural framework to generate realistic synthetic micro-Doppler signature dataset. The proposed network extracts environmental effects such as noise, multipath, and occlusions effects directly from the measurement data and transfers these features to the clean simulated signatures generated through our simulator called SimHumaLator. This results in more realistic-looking signatures qualitatively and quantitatively. We use these enhanced signatures to augment our measurement data and observe an improvement in the classification performance by 5% compared to no augmentation case.
Bocus, Mohammud Junaid; Li, Wenda; Vishwakarma, Shelly; Kou, Roget; Tang, Chong; Woodbridge, Karl; Craddock, Ian; McConville, Ryan; Santos-Rodriguez, Raul; Chetty, Kevin; others,
OPERAnet, a multimodal activity recognition dataset acquired from radio frequency and vision-based sensors Journal Article
In: Scientific data, vol. 9, no. 1, pp. 474, 2022.
Abstract | Links | BibTeX | Tags:
@article{bocus2022operanet,
title = {OPERAnet, a multimodal activity recognition dataset acquired from radio frequency and vision-based sensors},
author = {Mohammud Junaid Bocus and Wenda Li and Shelly Vishwakarma and Roget Kou and Chong Tang and Karl Woodbridge and Ian Craddock and Ryan McConville and Raul Santos-Rodriguez and Kevin Chetty and others},
url = {https://www.nature.com/articles/s41597-022-01573-2},
year = {2022},
date = {2022-01-01},
urldate = {2022-01-01},
journal = {Scientific data},
volume = {9},
number = {1},
pages = {474},
publisher = {Nature Publishing Group UK London},
abstract = {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.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
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.
Li, Wenda; Tang, Chong; Vishwakarma, Shelly; Woodbridge, Karl; Chetty, Kevin
Design of high-speed software defined radar with GPU accelerator Journal Article
In: IET Radar, Sonar & Navigation, vol. 16, no. 7, pp. 1083–1094, 2022.
Abstract | Links | BibTeX | Tags:
@article{li2022design,
title = {Design of high-speed software defined radar with GPU accelerator},
author = {Wenda Li and Chong Tang and Shelly Vishwakarma and Karl Woodbridge and Kevin Chetty},
url = {https://doi.org/10.1049/rsn2.12244},
year = {2022},
date = {2022-01-01},
urldate = {2022-01-01},
journal = {IET Radar, Sonar & Navigation},
volume = {16},
number = {7},
pages = {1083–1094},
publisher = {Wiley Online Library},
abstract = {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.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
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.
Tang, Chong; Li, Wenda; Vishwakarma, Shelly; Shi, Fangzhan; Julier, Simon; Chetty, Kevin
People counting using multistatic passive WiFi radar with a multi-input deep convolutional neural network Proceedings Article
In: Radar Sensor Technology XXVI, SPIE 2022.
Abstract | Links | BibTeX | Tags:
@inproceedings{tang2022people,
title = {People counting using multistatic passive WiFi radar with a multi-input deep convolutional neural network},
author = {Chong Tang and Wenda Li and Shelly Vishwakarma and Fangzhan Shi and Simon Julier and Kevin Chetty},
url = {https://discovery.ucl.ac.uk/id/eprint/10157000/},
year = {2022},
date = {2022-01-01},
urldate = {2022-01-01},
booktitle = {Radar Sensor Technology XXVI},
organization = {SPIE},
abstract = {Accurately counting numbers people is useful in many applications. Currently, camera-based systems assisted by computer vision and machine learning algorithms represent the state-of-the-art. However, they have limited coverage areas and are prone to blind spots, obscuration by walls, shadowing of individuals in crowds, and rely on optimal positioning and lighting conditions. Moreover, their ability to image people raises ethical and privacy concerns. In this paper we propose a distributed multistatic passive WiFi radar (PWR) consisting of 1 reference and 3 surveillance receivers, that can accurately count up to six test subjects using Doppler frequency shifts and intensity data from measured micro-Doppler (µ-Doppler) spectrograms. To build the person-counting processing model, we employ a multi-input convolutional neural network (MI-CNN). The results demonstrate a 96% counting accuracy for six subjects when data from all three surveillance channels are utilised.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Accurately counting numbers people is useful in many applications. Currently, camera-based systems assisted by computer vision and machine learning algorithms represent the state-of-the-art. However, they have limited coverage areas and are prone to blind spots, obscuration by walls, shadowing of individuals in crowds, and rely on optimal positioning and lighting conditions. Moreover, their ability to image people raises ethical and privacy concerns. In this paper we propose a distributed multistatic passive WiFi radar (PWR) consisting of 1 reference and 3 surveillance receivers, that can accurately count up to six test subjects using Doppler frequency shifts and intensity data from measured micro-Doppler (µ-Doppler) spectrograms. To build the person-counting processing model, we employ a multi-input convolutional neural network (MI-CNN). The results demonstrate a 96% counting accuracy for six subjects when data from all three surveillance channels are utilised.
Li, Wenda; Tang, Chong; Vishwakarma, Shelly; Shi, Fangzhan; Piechocki, Robert J; Chetty, Kevin
A High-Speed Multi-Purpose Software Defined Radar for Near-Field Applications Proceedings Article
In: 2022 IEEE Radar Conference (RadarConf22), pp. 1–6, IEEE 2022.
Abstract | Links | BibTeX | Tags:
@inproceedings{li2022high,
title = {A High-Speed Multi-Purpose Software Defined Radar for Near-Field Applications},
author = {Wenda Li and Chong Tang and Shelly Vishwakarma and Fangzhan Shi and Robert J Piechocki and Kevin Chetty},
url = {https://ieeexplore.ieee.org/abstract/document/9764260},
year = {2022},
date = {2022-01-01},
urldate = {2022-01-01},
booktitle = {2022 IEEE Radar Conference (RadarConf22)},
pages = {1–6},
organization = {IEEE},
abstract = {Software Defined Radar (SDRadar) is a unique radar system, where most of its processing, like filtering, correlation, signal generation etc. is performed by software. This means SDRadar can be flexibly deployed for different purposes and with a relative short development process. In this paper, we present a generic SDRadar system that can operate in different setups for near-field monitoring applications. Practical solutions for traditional limitations in SDRadar and high sampling rates are introduced, and its performance is demonstrated using a commercial universal software radio peripheral (USRP) device with four synchronized receiving channels and a maximum sampling rate of 100MHz. Additionally, a GPU accelerator has been implemented to deal with the high sampling rate. Three different setups have been tested to demonstrate the feasibility of the propose SDRadar system with distributed nodes, vertically positioned nodes and a miniature scenario. Recorded Doppler signatures have shown the proposed SDRadar can effectively capture the body and hand gestures. Such results can be used in a range of applications such as eHealth, human-machine interaction and indoor tracking.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Software Defined Radar (SDRadar) is a unique radar system, where most of its processing, like filtering, correlation, signal generation etc. is performed by software. This means SDRadar can be flexibly deployed for different purposes and with a relative short development process. In this paper, we present a generic SDRadar system that can operate in different setups for near-field monitoring applications. Practical solutions for traditional limitations in SDRadar and high sampling rates are introduced, and its performance is demonstrated using a commercial universal software radio peripheral (USRP) device with four synchronized receiving channels and a maximum sampling rate of 100MHz. Additionally, a GPU accelerator has been implemented to deal with the high sampling rate. Three different setups have been tested to demonstrate the feasibility of the propose SDRadar system with distributed nodes, vertically positioned nodes and a miniature scenario. Recorded Doppler signatures have shown the proposed SDRadar can effectively capture the body and hand gestures. Such results can be used in a range of applications such as eHealth, human-machine interaction and indoor tracking.
Shi, Fangzhan; Li, Wenda; Amiri, Amin; Vishwakarma, Shelly; Tang, Chong; Brennan, Paul; Chetty, Kevin
Pi-NIC: Indoor sensing using synchronized off-the-shelf wireless network interface cards and Raspberry Pis Proceedings Article
In: 2022 2nd IEEE International Symposium on Joint Communications & Sensing (JC&S), pp. 1–6, IEEE 2022.
Abstract | Links | BibTeX | Tags:
@inproceedings{shi2022pi,
title = {Pi-NIC: Indoor sensing using synchronized off-the-shelf wireless network interface cards and Raspberry Pis},
author = {Fangzhan Shi and Wenda Li and Amin Amiri and Shelly Vishwakarma and Chong Tang and Paul Brennan and Kevin Chetty},
url = {https://ieeexplore.ieee.org/abstract/document/9743512},
year = {2022},
date = {2022-01-01},
urldate = {2022-01-01},
booktitle = {2022 2nd IEEE International Symposium on Joint Communications & Sensing (JC&S)},
pages = {1–6},
organization = {IEEE},
abstract = {This paper presents an indoor joint communication and sensing system that consists of synchronized off-the-shelf wireless network interface cards (NIC) and Raspberry Pis. There exists a significant body of research that uses the channel state information (CSI) reported by wireless network interface cards for sensing, but only the amplitude and phase difference of the CSI between receiver antennas are processed. The raw phase of the CSI is contaminated by the carrier frequency offset, packet detection delay and other hardware imperfections, so it is too noisy to use. Our work introduces the raw phase of CSI into sensing by synchronizing the transmitter and receiver clocks to remove carrier frequency offset and using a new method to remove packet detection delay. We validate our design in a real-world scenario to detect breathing and walking and demonstrate that the raw phase of the CSI offers an evident improvement in Wi-Fi CSI-based sensing. Additionally, we are the first to use the Raspberry Pi and ATH9k wireless network interface card together for CSI data collection, which is cheap, portable and versatile.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
This paper presents an indoor joint communication and sensing system that consists of synchronized off-the-shelf wireless network interface cards (NIC) and Raspberry Pis. There exists a significant body of research that uses the channel state information (CSI) reported by wireless network interface cards for sensing, but only the amplitude and phase difference of the CSI between receiver antennas are processed. The raw phase of the CSI is contaminated by the carrier frequency offset, packet detection delay and other hardware imperfections, so it is too noisy to use. Our work introduces the raw phase of CSI into sensing by synchronizing the transmitter and receiver clocks to remove carrier frequency offset and using a new method to remove packet detection delay. We validate our design in a real-world scenario to detect breathing and walking and demonstrate that the raw phase of the CSI offers an evident improvement in Wi-Fi CSI-based sensing. Additionally, we are the first to use the Raspberry Pi and ATH9k wireless network interface card together for CSI data collection, which is cheap, portable and versatile.
2021
Vishwakarma, Shelly; Li, Wenda; Tang, Chong; Woodbridge, Karl; Adve, Raviraj; Chetty, Kevin
SimHumalator: An open-source end-to-end radar simulator for human activity recognition Journal Article
In: IEEE Aerospace and Electronic Systems Magazine, vol. 37, no. 3, pp. 6–22, 2021.
Abstract | Links | BibTeX | Tags:
@article{vishwakarma2021simhumalator,
title = {SimHumalator: An open-source end-to-end radar simulator for human activity recognition},
author = {Shelly Vishwakarma and Wenda Li and Chong Tang and Karl Woodbridge and Raviraj Adve and Kevin Chetty},
url = {https://ieeexplore.ieee.org/abstract/document/9664284},
year = {2021},
date = {2021-01-01},
urldate = {2021-01-01},
journal = {IEEE Aerospace and Electronic Systems Magazine},
volume = {37},
number = {3},
pages = {6–22},
publisher = {IEEE},
abstract = {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.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
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.
Li, Wenda; Vishwakarma, Shelly; Tang, Chong; Woodbridge, Karl; Piechocki, Robert J; Chetty, Kevin
Using RF transmissions from IoT devices for occupancy detection and activity Recognition Journal Article
In: IEEE Sensors Journal, vol. 22, no. 3, pp. 2484–2495, 2021.
Abstract | Links | BibTeX | Tags:
@article{li2021using,
title = {Using RF transmissions from IoT devices for occupancy detection and activity Recognition},
author = {Wenda Li and Shelly Vishwakarma and Chong Tang and Karl Woodbridge and Robert J Piechocki and Kevin Chetty},
url = {https://ieeexplore.ieee.org/abstract/document/9646920},
year = {2021},
date = {2021-01-01},
urldate = {2021-01-01},
journal = {IEEE Sensors Journal},
volume = {22},
number = {3},
pages = {2484–2495},
publisher = {IEEE},
abstract = {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).},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
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).
Tang, Chong; Li, Wenda; Vishwakarma, Shelly; Shi, Fangzhan; Julier, Simon; Chetty, Kevin
FMNet: Latent feature-wise mapping network for cleaning up noisy micro-Doppler spectrogram Journal Article
In: IEEE Transactions on Geoscience and Remote Sensing, vol. 60, pp. 1–12, 2021.
Abstract | Links | BibTeX | Tags:
@article{tang2021fmnet,
title = {FMNet: Latent feature-wise mapping network for cleaning up noisy micro-Doppler spectrogram},
author = {Chong Tang and Wenda Li and Shelly Vishwakarma and Fangzhan Shi and Simon Julier and Kevin Chetty},
url = {https://ieeexplore.ieee.org/abstract/document/9583945},
year = {2021},
date = {2021-01-01},
urldate = {2021-01-01},
journal = {IEEE Transactions on Geoscience and Remote Sensing},
volume = {60},
pages = {1–12},
publisher = {IEEE},
abstract = {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 ( μ -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.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
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 ( μ -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.
Vishwakarma, Shelly; Li, Wenda; Tang, Chong; Woodbridge, Karl; Adve, Raviraj; Chetty, Kevin
Neural style transfer enhanced training support for human activity recognition Journal Article
In: arXiv preprint arXiv:2107.12821, 2021.
Abstract | Links | BibTeX | Tags:
@article{vishwakarma2021neural,
title = {Neural style transfer enhanced training support for human activity recognition},
author = {Shelly Vishwakarma and Wenda Li and Chong Tang and Karl Woodbridge and Raviraj Adve and Kevin Chetty},
url = {
https://doi.org/10.48550/arXiv.2107.12821},
year = {2021},
date = {2021-01-01},
urldate = {2021-01-01},
journal = {arXiv preprint arXiv:2107.12821},
abstract = {This work presents an application of Integrated sensing and communication (ISAC) system for monitoring human activities directly related to healthcare. Real-time monitoring of humans can assist professionals in providing healthy living enabling technologies to ensure the health, safety, and well-being of people of all age groups. To enhance the human activity recognition performance of the ISAC system, we propose to use synthetic data generated through our human micro-Doppler simulator, SimHumalator to augment our limited measurement data. We generate a more realistic micro-Doppler signature dataset using a style-transfer neural network. The proposed network extracts environmental effects such as noise, multipath, and occlusions effects directly from the measurement data and transfers these features to our clean simulated signatures. This results in more realistic-looking signatures qualitatively and quantitatively. We use these enhanced signatures to augment our measurement data and observe an improvement in the classification performance by 5% compared to no augmentation case. Further, we benchmark the data augmentation performance of the style transferred signatures with three other synthetic datasets -- clean simulated spectrograms (no environmental effects), simulated data with added AWGN noise, and simulated data with GAN generated noise. The results indicate that style transferred simulated signatures well captures environmental factors more than any other synthetic dataset.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
This work presents an application of Integrated sensing and communication (ISAC) system for monitoring human activities directly related to healthcare. Real-time monitoring of humans can assist professionals in providing healthy living enabling technologies to ensure the health, safety, and well-being of people of all age groups. To enhance the human activity recognition performance of the ISAC system, we propose to use synthetic data generated through our human micro-Doppler simulator, SimHumalator to augment our limited measurement data. We generate a more realistic micro-Doppler signature dataset using a style-transfer neural network. The proposed network extracts environmental effects such as noise, multipath, and occlusions effects directly from the measurement data and transfers these features to our clean simulated signatures. This results in more realistic-looking signatures qualitatively and quantitatively. We use these enhanced signatures to augment our measurement data and observe an improvement in the classification performance by 5% compared to no augmentation case. Further, we benchmark the data augmentation performance of the style transferred signatures with three other synthetic datasets — clean simulated spectrograms (no environmental effects), simulated data with added AWGN noise, and simulated data with GAN generated noise. The results indicate that style transferred simulated signatures well captures environmental factors more than any other synthetic dataset.
Vishwakarma, Shelly; Tang, Chong; Li, Wenda; Woodbridge, Karl; Adve, Raviraj; Chetty, Kevin
Gan based noise generation to aid activity recognition when augmenting measured wifi radar data with simulations Proceedings Article
In: 2021 IEEE International Conference on Communications Workshops (ICC Workshops), pp. 1–6, IEEE 2021.
Abstract | Links | BibTeX | Tags:
@inproceedings{vishwakarma2021gan,
title = {Gan based noise generation to aid activity recognition when augmenting measured wifi radar data with simulations},
author = {Shelly Vishwakarma and Chong Tang and Wenda Li and Karl Woodbridge and Raviraj Adve and Kevin Chetty},
url = {https://ieeexplore.ieee.org/abstract/document/9473900},
year = {2021},
date = {2021-01-01},
urldate = {2021-01-01},
booktitle = {2021 IEEE International Conference on Communications Workshops (ICC Workshops)},
pages = {1–6},
organization = {IEEE},
abstract = {This work considers the use of a passive WiFi radar (PWR) to monitor human activities. Real-time uncooperative monitoring of people has numerous applications ranging from smart cities and transport to IoT and security. In e-healthcare, PWR technology could be used for ambient assisted living and early detection of chronic health conditions. Large training datasets could drive forward machine-learning-focused research in the above applications. However, generating and labeling large volumes of high-quality, diverse radar datasets is an onerous task. Therefore, we present an open-source motion capture data-driven simulation tool, SimHumalator, that can generate large volumes of human micro-Doppler radar data at multiple IEEE WiFi standards(IEEE 802.11g, n, and ad). We qualitatively compare the micro-Doppler signatures generated through SimHumalator with the measured signatures. To create a more realistic training dataset, we artificially add noise to our clean simulated spectrograms. A noise distribution is directly learned from real radar measurements using a Generative Adversarial Network (GAN). We observe improvements in the classification performances between 3 to 8%. Our results suggest that simulation data can be used to make adequate training data when the available measurement training support is low.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
This work considers the use of a passive WiFi radar (PWR) to monitor human activities. Real-time uncooperative monitoring of people has numerous applications ranging from smart cities and transport to IoT and security. In e-healthcare, PWR technology could be used for ambient assisted living and early detection of chronic health conditions. Large training datasets could drive forward machine-learning-focused research in the above applications. However, generating and labeling large volumes of high-quality, diverse radar datasets is an onerous task. Therefore, we present an open-source motion capture data-driven simulation tool, SimHumalator, that can generate large volumes of human micro-Doppler radar data at multiple IEEE WiFi standards(IEEE 802.11g, n, and ad). We qualitatively compare the micro-Doppler signatures generated through SimHumalator with the measured signatures. To create a more realistic training dataset, we artificially add noise to our clean simulated spectrograms. A noise distribution is directly learned from real radar measurements using a Generative Adversarial Network (GAN). We observe improvements in the classification performances between 3 to 8%. Our results suggest that simulation data can be used to make adequate training data when the available measurement training support is low.
Dhulashia, Dilan; Ritchie, Matthew; Vishwakarma, Shelly; Chetty, Kevin
Human micro-Doppler signature classification in the presence of a selection of jamming signals Proceedings Article
In: 2021 IEEE Radar Conference (RadarConf21), pp. 1–6, IEEE 2021.
Abstract | Links | BibTeX | Tags:
@inproceedings{dhulashia2021human,
title = {Human micro-Doppler signature classification in the presence of a selection of jamming signals},
author = {Dilan Dhulashia and Matthew Ritchie and Shelly Vishwakarma and Kevin Chetty},
url = {https://ieeexplore.ieee.org/abstract/document/9455278},
year = {2021},
date = {2021-01-01},
urldate = {2021-01-01},
booktitle = {2021 IEEE Radar Conference (RadarConf21)},
pages = {1–6},
organization = {IEEE},
abstract = {This work investigates the degradation effects of four distinct jamming signal styles on human micro-Doppler signatures by examining the ability of a linear discriminant classifier to accurately distinguish signatures collected using a simulated frequency modulated continuous wave (FMCW) radar which have been injected with jamming. Misclassification dependence on jamming signal power for each jamming style is presented along with the nature of misclassifications.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
This work investigates the degradation effects of four distinct jamming signal styles on human micro-Doppler signatures by examining the ability of a linear discriminant classifier to accurately distinguish signatures collected using a simulated frequency modulated continuous wave (FMCW) radar which have been injected with jamming. Misclassification dependence on jamming signal power for each jamming style is presented along with the nature of misclassifications.
Tang, Chong; Vishwakarma, Shelly; Li, Wenda; Adve, Raviraj; Julier, Simon; Chetty, Kevin
Augmenting experimental data with simulations to improve activity classification in healthcare monitoring Proceedings Article
In: 2021 IEEE radar conference (RadarConf21), pp. 1–6, IEEE 2021.
Abstract | Links | BibTeX | Tags:
@inproceedings{tang2021augmenting,
title = {Augmenting experimental data with simulations to improve activity classification in healthcare monitoring},
author = {Chong Tang and Shelly Vishwakarma and Wenda Li and Raviraj Adve and Simon Julier and Kevin Chetty},
url = {https://ieeexplore.ieee.org/abstract/document/9455314},
year = {2021},
date = {2021-01-01},
urldate = {2021-01-01},
booktitle = {2021 IEEE radar conference (RadarConf21)},
pages = {1–6},
organization = {IEEE},
abstract = {Human micro-Doppler signatures in most passive WiFi radar (PWR) scenarios are captured through real-world measurements using various hardware platforms. However, gathering large volumes of high quality and diverse real radar datasets has always been an expensive and laborious task. This work presents an open-source motion capture data-driven simulation tool SimHumalator that is able to generate human micro-Doppler radar data in PWR scenarios. We qualitatively compare the micro-Doppler signatures generated through SimHumalator with the measured real signatures. Here, we present the use of SimHumalator to simulate a set of human actions. We demonstrate that augmenting a measurement database with simulated data, using SimHumalator, results in an 8% improvement in classification accuracy. Our results suggest that simulation data can be used to augment experimental datasets of limited volume to address the cold-start problem typically encountered in radar research.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Human micro-Doppler signatures in most passive WiFi radar (PWR) scenarios are captured through real-world measurements using various hardware platforms. However, gathering large volumes of high quality and diverse real radar datasets has always been an expensive and laborious task. This work presents an open-source motion capture data-driven simulation tool SimHumalator that is able to generate human micro-Doppler radar data in PWR scenarios. We qualitatively compare the micro-Doppler signatures generated through SimHumalator with the measured real signatures. Here, we present the use of SimHumalator to simulate a set of human actions. We demonstrate that augmenting a measurement database with simulated data, using SimHumalator, results in an 8% improvement in classification accuracy. Our results suggest that simulation data can be used to augment experimental datasets of limited volume to address the cold-start problem typically encountered in radar research.
Tang, Chong; Li, Wenda; Vishwakarma, Shelly; Woodbridge, Karl; Julier, Simon; Chetty, Kevin
Learning from natural noise to denoise micro-doppler spectrogram Journal Article
In: arXiv preprint arXiv:2102.06887, 2021.
Abstract | Links | BibTeX | Tags:
@article{tang2021learning,
title = {Learning from natural noise to denoise micro-doppler spectrogram},
author = {Chong Tang and Wenda Li and Shelly Vishwakarma and Karl Woodbridge and Simon Julier and Kevin Chetty},
url = {
https://doi.org/10.48550/arXiv.2102.06887},
year = {2021},
date = {2021-01-01},
urldate = {2021-01-01},
journal = {arXiv preprint arXiv:2102.06887},
abstract = {Micro-Doppler analysis has become increasingly popular in recent years owning to the ability of the technique to enhance classification strategies. Applications include recognising everyday human activities, distinguishing drone from birds, and identifying different types of vehicles. However, noisy time-frequency spectrograms can significantly affect the performance of the classifier and must be tackled using appropriate denoising algorithms. In recent years, deep learning algorithms have spawned many deep neural network-based denoising algorithms. For these methods, noise modelling is the most important part and is used to assist in training. In this paper, we decompose the problem and propose a novel denoising scheme: first, a Generative Adversarial Network (GAN) is used to learn the noise distribution and correlation from the real-world environment; then, a simulator is used to generate clean Micro-Doppler spectrograms; finally, the generated noise and clean simulation data are combined as the training data to train a Convolutional Neural Network (CNN) denoiser. In experiments, we qualitatively and quantitatively analyzed this procedure on both simulation and measurement data. Besides, the idea of learning from natural noise can be applied well to other existing frameworks and demonstrate greater performance than other noise models.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Micro-Doppler analysis has become increasingly popular in recent years owning to the ability of the technique to enhance classification strategies. Applications include recognising everyday human activities, distinguishing drone from birds, and identifying different types of vehicles. However, noisy time-frequency spectrograms can significantly affect the performance of the classifier and must be tackled using appropriate denoising algorithms. In recent years, deep learning algorithms have spawned many deep neural network-based denoising algorithms. For these methods, noise modelling is the most important part and is used to assist in training. In this paper, we decompose the problem and propose a novel denoising scheme: first, a Generative Adversarial Network (GAN) is used to learn the noise distribution and correlation from the real-world environment; then, a simulator is used to generate clean Micro-Doppler spectrograms; finally, the generated noise and clean simulation data are combined as the training data to train a Convolutional Neural Network (CNN) denoiser. In experiments, we qualitatively and quantitatively analyzed this procedure on both simulation and measurement data. Besides, the idea of learning from natural noise can be applied well to other existing frameworks and demonstrate greater performance than other noise models.
Ram, Shobha Sundar; Vishwakarma, Shelly; Sneh, Akanksha; Yasmeen, Kainat
Sparsity-based autoencoders for denoising cluttered radar signatures Journal Article
In: IET Radar, Sonar & Navigation, vol. 15, no. 8, pp. 915–931, 2021.
Abstract | Links | BibTeX | Tags:
@article{ram2021sparsity,
title = {Sparsity-based autoencoders for denoising cluttered radar signatures},
author = {Shobha Sundar Ram and Shelly Vishwakarma and Akanksha Sneh and Kainat Yasmeen},
url = {https://doi.org/10.1049/rsn2.12065},
year = {2021},
date = {2021-01-01},
urldate = {2021-01-01},
journal = {IET Radar, Sonar & Navigation},
volume = {15},
number = {8},
pages = {915–931},
publisher = {Wiley Online Library},
abstract = {Narrowband and broadband indoor radar images significantly deteriorate in the presence of target-dependent and target-independent static and dynamic clutter arising from walls. A stacked and sparse denoising autoencoder (StackedSDAE) is proposed for mitigating the wall clutter in indoor radar images. The algorithm relies on the availability of clean images and the corresponding noisy images during training and requires no additional information regarding the wall characteristics. The algorithm is evaluated on simulated Doppler-time spectrograms and high-range resolution profiles generated for diverse radar frequencies and wall characteristics in around-the-corner radar (ACR) scenarios. Additional experiments are performed on range-enhanced frontal images generated from measurements gathered from a wideband radio frequency imaging sensor. The results from the experiments show that the StackedSDAE successfully reconstructs images that closely resemble those that would be obtained in free space conditions. Furthermore, the incorporation of sparsity and depth in the hidden layer representations within the autoencoder makes the algorithm more robust to low signal-to-noise ratio (SNR) and label mismatch between clean and corrupt data during training than the conventional single-layer DAE. For example, the denoised ACR signatures show a structural similarity above 0.75 to clean free space images at SNR of −10 dB and label mismatch error of 50%.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Narrowband and broadband indoor radar images significantly deteriorate in the presence of target-dependent and target-independent static and dynamic clutter arising from walls. A stacked and sparse denoising autoencoder (StackedSDAE) is proposed for mitigating the wall clutter in indoor radar images. The algorithm relies on the availability of clean images and the corresponding noisy images during training and requires no additional information regarding the wall characteristics. The algorithm is evaluated on simulated Doppler-time spectrograms and high-range resolution profiles generated for diverse radar frequencies and wall characteristics in around-the-corner radar (ACR) scenarios. Additional experiments are performed on range-enhanced frontal images generated from measurements gathered from a wideband radio frequency imaging sensor. The results from the experiments show that the StackedSDAE successfully reconstructs images that closely resemble those that would be obtained in free space conditions. Furthermore, the incorporation of sparsity and depth in the hidden layer representations within the autoencoder makes the algorithm more robust to low signal-to-noise ratio (SNR) and label mismatch between clean and corrupt data during training than the conventional single-layer DAE. For example, the denoised ACR signatures show a structural similarity above 0.75 to clean free space images at SNR of −10 dB and label mismatch error of 50%.
2020
Tran, Kimberly T; Griffin, Lewis D; Chetty, Kevin; Vishwakarma, Shelly
Transfer learning from audio deep learning models for micro-Doppler activity recognition Proceedings Article
In: 2020 IEEE International Radar Conference (RADAR), pp. 584–589, IEEE 2020.
Abstract | Links | BibTeX | Tags:
@inproceedings{tran2020transfer,
title = {Transfer learning from audio deep learning models for micro-Doppler activity recognition},
author = {Kimberly T Tran and Lewis D Griffin and Kevin Chetty and Shelly Vishwakarma},
url = {https://ieeexplore.ieee.org/abstract/document/9114643},
year = {2020},
date = {2020-01-01},
urldate = {2020-01-01},
booktitle = {2020 IEEE International Radar Conference (RADAR)},
pages = {584–589},
organization = {IEEE},
abstract = {This paper presents a mechanism to transform radio frequency micro-Doppler signatures into a pseudo-audio representation, which results in significant improvements in transfer learning from a deep learning model trained on audio. We also demonstrate that transfer learning from a deep learning model trained on audio is more effective than transfer learning from a model trained on images, which suggests machine learning methods used to analyse audio can be leveraged for micro-Doppler. Finally, we utilise an occlusion method to gain an insight into how the deep learning model interprets the micro-Doppler signatures and the subsequent pseudo-audio representations.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
This paper presents a mechanism to transform radio frequency micro-Doppler signatures into a pseudo-audio representation, which results in significant improvements in transfer learning from a deep learning model trained on audio. We also demonstrate that transfer learning from a deep learning model trained on audio is more effective than transfer learning from a model trained on images, which suggests machine learning methods used to analyse audio can be leveraged for micro-Doppler. Finally, we utilise an occlusion method to gain an insight into how the deep learning model interprets the micro-Doppler signatures and the subsequent pseudo-audio representations.
Vishwakarma, Shelly; Rafiq, Aaquib; Ram, Shobha Sundar
Micro-Doppler signatures of dynamic humans from around the corner radar Proceedings Article
In: 2020 IEEE International Radar Conference (RADAR), pp. 169–174, IEEE 2020.
Abstract | Links | BibTeX | Tags:
@inproceedings{vishwakarma2020micro,
title = {Micro-Doppler signatures of dynamic humans from around the corner radar},
author = {Shelly Vishwakarma and Aaquib Rafiq and Shobha Sundar Ram},
url = {https://ieeexplore.ieee.org/abstract/document/9114675},
year = {2020},
date = {2020-01-01},
urldate = {2020-01-01},
booktitle = {2020 IEEE International Radar Conference (RADAR)},
pages = {169–174},
organization = {IEEE},
abstract = {Recent studies have demonstrated the possibility of sensing dynamic targets around the corners with no direct signal in line-of-sight with respect to the radar. These works have mostly focused on the detection of targets around the corner on the basis of multipath scattering from lateral walls. However, strong specular multipath returns are only obtained for highly conductive walls or at high carrier frequencies. There is minimal research effort into using the existing indoor radar hardware at much lower carrier frequencies for around the corner sensing of targets. In this paper, we have performed a detailed experimental analysis, including both simulations and measurements, of the effect of wall parameters and carrier frequency on the around the corner micro-Doppler signatures of dynamic humans. Our results demonstrate that in real world scenarios where walls are lossy, target micro-Dopplers are weak and distorted by multipath scattering at high carrier frequencies and are sensed only very near the radar. At lower carrier frequencies, the targets are sensed at greater distances and the micro-Dopplers are not significantly distorted by multipath since the signals mostly travel along the direct path through the wall.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Recent studies have demonstrated the possibility of sensing dynamic targets around the corners with no direct signal in line-of-sight with respect to the radar. These works have mostly focused on the detection of targets around the corner on the basis of multipath scattering from lateral walls. However, strong specular multipath returns are only obtained for highly conductive walls or at high carrier frequencies. There is minimal research effort into using the existing indoor radar hardware at much lower carrier frequencies for around the corner sensing of targets. In this paper, we have performed a detailed experimental analysis, including both simulations and measurements, of the effect of wall parameters and carrier frequency on the around the corner micro-Doppler signatures of dynamic humans. Our results demonstrate that in real world scenarios where walls are lossy, target micro-Dopplers are weak and distorted by multipath scattering at high carrier frequencies and are sensed only very near the radar. At lower carrier frequencies, the targets are sensed at greater distances and the micro-Dopplers are not significantly distorted by multipath since the signals mostly travel along the direct path through the wall.
Li, Wenda; Bocus, Mohammud Junaid; Tang, Chong; Vishwakarma, Shelly; Piechocki, Robert J; Woodbridge, Karl; Chetty, Kevin
A taxonomy of WiFi sensing: CSI vs passive WiFi radar Proceedings Article
In: 2020 IEEE Globecom Workshops (GC Wkshps, pp. 1–6, IEEE 2020.
Abstract | Links | BibTeX | Tags:
@inproceedings{li2020taxonomy,
title = {A taxonomy of WiFi sensing: CSI vs passive WiFi radar},
author = {Wenda Li and Mohammud Junaid Bocus and Chong Tang and Shelly Vishwakarma and Robert J Piechocki and Karl Woodbridge and Kevin Chetty},
url = {https://ieeexplore.ieee.org/abstract/document/9367546},
year = {2020},
date = {2020-01-01},
urldate = {2020-01-01},
booktitle = {2020 IEEE Globecom Workshops (GC Wkshps},
pages = {1–6},
organization = {IEEE},
abstract = {WiFi sensing has shown promising potentials in a number of applications such as healthcare, smart transportation and home automation. Human activity recognition without the use of any cooperative device such as phones or wearable technologies can be achieved by two WiFi based approaches: Channel State Information (CSI) and Passive WiFi Radar (PWR). CSI systems rely directly on WiFi as a communications system, whilst PWR treats the WiFi signal as an illuminator for use in a radar signal processing. However, there has not been a comprehensive comparative study on the similarities and differences between the two systems. To examine the performance of both systems we implement two hardware platforms for CSI and PWR, and use them concurrently to capture the human movements. In this paper, we present Doppler measurements from the two systems and compare their performance using a dataset obtained from five subjects undergoing six activity classes. It is observed that both systems have very different Doppler signatures, and are sensitive to the transmitter-target-receiver geometries. CSI has a better performance in Line-of-Sight (LoS) configurations, whereas PWR has better performance in bistatic configurations where the WiFi access point and radar receiver are spatially separated. It is envisioned that a more robust system should leverage strengths of both the CSI and PWR systems jointly to maximize their benefits in wireless sensing.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
WiFi sensing has shown promising potentials in a number of applications such as healthcare, smart transportation and home automation. Human activity recognition without the use of any cooperative device such as phones or wearable technologies can be achieved by two WiFi based approaches: Channel State Information (CSI) and Passive WiFi Radar (PWR). CSI systems rely directly on WiFi as a communications system, whilst PWR treats the WiFi signal as an illuminator for use in a radar signal processing. However, there has not been a comprehensive comparative study on the similarities and differences between the two systems. To examine the performance of both systems we implement two hardware platforms for CSI and PWR, and use them concurrently to capture the human movements. In this paper, we present Doppler measurements from the two systems and compare their performance using a dataset obtained from five subjects undergoing six activity classes. It is observed that both systems have very different Doppler signatures, and are sensitive to the transmitter-target-receiver geometries. CSI has a better performance in Line-of-Sight (LoS) configurations, whereas PWR has better performance in bistatic configurations where the WiFi access point and radar receiver are spatially separated. It is envisioned that a more robust system should leverage strengths of both the CSI and PWR systems jointly to maximize their benefits in wireless sensing.
Tang, Chong; Li, Wenda; Vishwakarma, Shelly; Chetty, Kevin; Julier, Simon; Woodbridge, Karl
Occupancy detection and people counting using wifi passive radar Proceedings Article
In: 2020 IEEE Radar Conference (RadarConf20), pp. 1–6, IEEE 2020.
Abstract | Links | BibTeX | Tags:
@inproceedings{tang2020occupancy,
title = {Occupancy detection and people counting using wifi passive radar},
author = {Chong Tang and Wenda Li and Shelly Vishwakarma and Kevin Chetty and Simon Julier and Karl Woodbridge},
url = {https://ieeexplore.ieee.org/abstract/document/9266493},
year = {2020},
date = {2020-01-01},
urldate = {2020-01-01},
booktitle = {2020 IEEE Radar Conference (RadarConf20)},
pages = {1–6},
organization = {IEEE},
abstract = {Occupancy detection and people counting technologies have important uses in many scenarios ranging from management of human resources, optimising energy use in intelligent buildings and improving public services in future smart cities. Wi-Fi based sensing approaches for these applications have attracted significant attention in recent years because of their ubiquitous nature, and ability to preserve the privacy of individuals being counted. In this paper, we present a Passive Wi-Fi Radar (PWR) technique for occupancy detection and people counting. Unlike systems which exploit the Wi-Fi Received Signal Strength (RSS) and Channel State Information (CSI), PWR systems can directly be applied in any environment covered by an existing WiFi local area network without special modifications to the Wi-Fi access point. Specifically, we apply Cross Ambiguity Function (CAF) processing to generate Range-Doppler maps, then we use Time-Frequency transforms to generate Doppler spectrograms, and finally employ a CLEAN algorithm to remove the direct signal interference. A Convolutional Neural Network (CNN) and sliding-window based feature selection scheme is then used for classification. Experimental results collected from a typical office environment are used to validate the proposed PWR system for accurately determining room occupancy, and correctly predict the number of people when using four test subjects in experimental measurements.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Occupancy detection and people counting technologies have important uses in many scenarios ranging from management of human resources, optimising energy use in intelligent buildings and improving public services in future smart cities. Wi-Fi based sensing approaches for these applications have attracted significant attention in recent years because of their ubiquitous nature, and ability to preserve the privacy of individuals being counted. In this paper, we present a Passive Wi-Fi Radar (PWR) technique for occupancy detection and people counting. Unlike systems which exploit the Wi-Fi Received Signal Strength (RSS) and Channel State Information (CSI), PWR systems can directly be applied in any environment covered by an existing WiFi local area network without special modifications to the Wi-Fi access point. Specifically, we apply Cross Ambiguity Function (CAF) processing to generate Range-Doppler maps, then we use Time-Frequency transforms to generate Doppler spectrograms, and finally employ a CLEAN algorithm to remove the direct signal interference. A Convolutional Neural Network (CNN) and sliding-window based feature selection scheme is then used for classification. Experimental results collected from a typical office environment are used to validate the proposed PWR system for accurately determining room occupancy, and correctly predict the number of people when using four test subjects in experimental measurements.
Duggal, Gaurav; Vishwakarma, Shelly; Mishra, Kumar Vijay; Ram, Shobha Sundar
Doppler-resilient 802.11 ad-based ultrashort range automotive joint radar-communications system Journal Article
In: IEEE Transactions on Aerospace and Electronic Systems, vol. 56, no. 5, pp. 4035–4048, 2020.
Abstract | Links | BibTeX | Tags:
@article{duggal2020doppler,
title = {Doppler-resilient 802.11 ad-based ultrashort range automotive joint radar-communications system},
author = {Gaurav Duggal and Shelly Vishwakarma and Kumar Vijay Mishra and Shobha Sundar Ram},
url = {https://ieeexplore.ieee.org/abstract/document/9086030},
year = {2020},
date = {2020-01-01},
urldate = {2020-01-01},
journal = {IEEE Transactions on Aerospace and Electronic Systems},
volume = {56},
number = {5},
pages = {4035–4048},
publisher = {IEEE},
abstract = {We present an ultrashort range IEEE 802.11ad-based automotive joint radar-communications (JRC) framework, wherein we improve the radar's Doppler resilience by incorporating Prouhet-Thue-Morse sequences in the preamble. The proposed processing reveals detailed microfeatures of common automotive objects verified through extended scattering center models of animated pedestrian, bicycle, and car targets. Numerical experiments demonstrate 2.5% reduction in the probability of false alarm at low signal-to-noise-ratios and improvement in the peak-to-sidelobe level dynamic range up to Doppler velocities of ±144 km/h over conventional 802.11ad JRC.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
We present an ultrashort range IEEE 802.11ad-based automotive joint radar-communications (JRC) framework, wherein we improve the radar's Doppler resilience by incorporating Prouhet-Thue-Morse sequences in the preamble. The proposed processing reveals detailed microfeatures of common automotive objects verified through extended scattering center models of animated pedestrian, bicycle, and car targets. Numerical experiments demonstrate 2.5% reduction in the probability of false alarm at low signal-to-noise-ratios and improvement in the peak-to-sidelobe level dynamic range up to Doppler velocities of ±144 km/h over conventional 802.11ad JRC.
Vishwakarma, Shelly; Ram, Shobha Sundar
Mitigation of through-wall distortions of frontal radar images using denoising autoencoders Journal Article
In: IEEE transactions on geoscience and remote sensing, vol. 58, no. 9, pp. 6650–6663, 2020.
Abstract | Links | BibTeX | Tags:
@article{vishwakarma2020mitigation,
title = {Mitigation of through-wall distortions of frontal radar images using denoising autoencoders},
author = {Shelly Vishwakarma and Shobha Sundar Ram},
url = {https://ieeexplore.ieee.org/abstract/document/9040872},
year = {2020},
date = {2020-01-01},
urldate = {2020-01-01},
journal = {IEEE transactions on geoscience and remote sensing},
volume = {58},
number = {9},
pages = {6650–6663},
publisher = {IEEE},
abstract = {Radar images of humans and other concealed objects are considerably distorted by attenuation, refraction, and multipath clutter in indoor through-wall environments. Although several methods have been proposed for removing target-independent static and dynamic clutter, there still remain considerable challenges in mitigating target-dependent clutter especially when the knowledge of the exact propagation characteristics or analytical framework is unavailable. In this article, we focus on mitigating wall effects using a machine learning-based solution-denoising autoencoders-that does not require prior information of the wall parameters or room geometry. Instead, the method relies on the availability of a large volume of training radar images gathered in through-wall conditions and the corresponding clean images captured in line-of-sight conditions. During the training phase, the autoencoder learns how to denoise the corrupted through-wall images in order to resemble the free space images. We have validated the performance of the proposed solution for both static and dynamic human subjects. The frontal radar images of static targets are obtained by processing wideband planar array measurement data with 2-D array and range processing. The frontal radar images of dynamic targets are simulated using narrowband planar array data processed with 2-D array and Doppler processing. In both simulation and measurement processes, we incorporate considerable diversity in the target and propagation conditions. Our experimental results, from both simulation and measurement data, show that the denoised images are considerably more similar to the free-space images when compared to the original through-wall images.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Radar images of humans and other concealed objects are considerably distorted by attenuation, refraction, and multipath clutter in indoor through-wall environments. Although several methods have been proposed for removing target-independent static and dynamic clutter, there still remain considerable challenges in mitigating target-dependent clutter especially when the knowledge of the exact propagation characteristics or analytical framework is unavailable. In this article, we focus on mitigating wall effects using a machine learning-based solution-denoising autoencoders-that does not require prior information of the wall parameters or room geometry. Instead, the method relies on the availability of a large volume of training radar images gathered in through-wall conditions and the corresponding clean images captured in line-of-sight conditions. During the training phase, the autoencoder learns how to denoise the corrupted through-wall images in order to resemble the free space images. We have validated the performance of the proposed solution for both static and dynamic human subjects. The frontal radar images of static targets are obtained by processing wideband planar array measurement data with 2-D array and range processing. The frontal radar images of dynamic targets are simulated using narrowband planar array data processed with 2-D array and Doppler processing. In both simulation and measurement processes, we incorporate considerable diversity in the target and propagation conditions. Our experimental results, from both simulation and measurement data, show that the denoised images are considerably more similar to the free-space images when compared to the original through-wall images.
Vishwakarma, Shelly; Ram, Shobha Sundar
IIIT-Delhi, 2020.
@phdthesis{vishwakarma2020learning,
title = {Learning algorithms for micro-doppler radar based detection, classification and imaging of humans in indoor environments},
author = {Shelly Vishwakarma and Shobha Sundar Ram},
url = {https://repository.iiitd.edu.in/jspui/handle/123456789/800},
year = {2020},
date = {2020-01-01},
urldate = {2020-01-01},
school = {IIIT-Delhi},
keywords = {},
pubstate = {published},
tppubtype = {phdthesis}
}
2019
Vishwakarma, Shelly; Pandey, Neeraj; Ram, Shobha Sundar
Clutter mitigation in range enhanced radar images using sparsity based denoising autoencoders Proceedings Article
In: 2019 International Radar Conference (RADAR), pp. 1–6, IEEE 2019.
Abstract | Links | BibTeX | Tags:
@inproceedings{vishwakarma2019clutter,
title = {Clutter mitigation in range enhanced radar images using sparsity based denoising autoencoders},
author = {Shelly Vishwakarma and Neeraj Pandey and Shobha Sundar Ram},
url = {https://ieeexplore.ieee.org/abstract/document/9079020},
year = {2019},
date = {2019-01-01},
urldate = {2019-01-01},
booktitle = {2019 International Radar Conference (RADAR)},
pages = {1–6},
organization = {IEEE},
abstract = {Complex propagation channels in indoor environments introduce considerable clutter and distortions to frontal radar images. Autoencoders, when provided with corrupted and corresponding clean images during training, can learn how to mitigate clutter in these images during testing. The main advantage is that an analytical framework of the channel is not required. On the other hand, the algorithm relies on the availability of a large database of correctly labelled training data - the target scenarios in cluttered environments must be exactly replicated in free space. We propose to augment the conventional denoising autoencoder with sparsity based multiple layer representations to handle the labelling mismatch errors. We hypothesize that the deeper representations with sparsity constraints will enable extraction of more fundamental features of the images. Our radar data consists of measurements made with a wideband RF imaging sensor called the Walabot. Our algorithm improves upon the conventional autoencoder in terms of performance with respect to signal to clutter and signal to noise ratios, labelling mismatch errors and computational time during testing.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Complex propagation channels in indoor environments introduce considerable clutter and distortions to frontal radar images. Autoencoders, when provided with corrupted and corresponding clean images during training, can learn how to mitigate clutter in these images during testing. The main advantage is that an analytical framework of the channel is not required. On the other hand, the algorithm relies on the availability of a large database of correctly labelled training data – the target scenarios in cluttered environments must be exactly replicated in free space. We propose to augment the conventional denoising autoencoder with sparsity based multiple layer representations to handle the labelling mismatch errors. We hypothesize that the deeper representations with sparsity constraints will enable extraction of more fundamental features of the images. Our radar data consists of measurements made with a wideband RF imaging sensor called the Walabot. Our algorithm improves upon the conventional autoencoder in terms of performance with respect to signal to clutter and signal to noise ratios, labelling mismatch errors and computational time during testing.
2018
Vishwakarma, Shelly; Ram, Shobha Sundar
Dictionary learning with low computational complexity for classification of human micro-Dopplers across multiple carrier frequencies Journal Article
In: IEEE Access, vol. 6, pp. 29793–29805, 2018.
Abstract | Links | BibTeX | Tags:
@article{vishwakarma2018dictionary,
title = {Dictionary learning with low computational complexity for classification of human micro-Dopplers across multiple carrier frequencies},
author = {Shelly Vishwakarma and Shobha Sundar Ram},
url = {https://ieeexplore.ieee.org/abstract/document/8375093},
year = {2018},
date = {2018-01-01},
urldate = {2018-01-01},
journal = {IEEE Access},
volume = {6},
pages = {29793–29805},
publisher = {IEEE},
abstract = {Recently, several machine learning algorithms have been applied for classifying micro-Doppler signatures from different human motions. However, these algorithms must demonstrate versatility in handling diversity in test and training data to be used for real-life scenarios. For example, situations may arise where the propagation channel or the presence of interference sources in the test site will permit only specific frequency bands of radar operation. These bands may differ from those used previously while training. In this paper, we examine the performances of three sparsity driven dictionary learning algorithms—synthesis, deep, and analysis—for learning unique features extracted from training data gathered across multiple carrier frequencies. These features are subsequently used for classifying test data from another distinct carrier frequency. Our experimental results, from measurement data, show that the dictionary learning algorithms are capable of extracting meaningful representations of the micro-Dopplers despite the rich frequency diversity in the data. In particular, the deep dictionary learning algorithm yields a high classification accuracy of 91% with a very low computational time for testing.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Recently, several machine learning algorithms have been applied for classifying micro-Doppler signatures from different human motions. However, these algorithms must demonstrate versatility in handling diversity in test and training data to be used for real-life scenarios. For example, situations may arise where the propagation channel or the presence of interference sources in the test site will permit only specific frequency bands of radar operation. These bands may differ from those used previously while training. In this paper, we examine the performances of three sparsity driven dictionary learning algorithms—synthesis, deep, and analysis—for learning unique features extracted from training data gathered across multiple carrier frequencies. These features are subsequently used for classifying test data from another distinct carrier frequency. Our experimental results, from measurement data, show that the dictionary learning algorithms are capable of extracting meaningful representations of the micro-Dopplers despite the rich frequency diversity in the data. In particular, the deep dictionary learning algorithm yields a high classification accuracy of 91% with a very low computational time for testing.
Vishwakarma, Shelly; Ummalaneni, Vahini; Iqbal, Muhammad Shoaib; Majumdar, Angshul; Ram, Shobha Sundar
Mitigation of through-wall interference in radar images using denoising autoencoders Proceedings Article
In: 2018 IEEE Radar Conference (RadarConf18), pp. 1543–1548, IEEE 2018.
Abstract | Links | BibTeX | Tags:
@inproceedings{vishwakarma2018mitigation,
title = {Mitigation of through-wall interference in radar images using denoising autoencoders},
author = {Shelly Vishwakarma and Vahini Ummalaneni and Muhammad Shoaib Iqbal and Angshul Majumdar and Shobha Sundar Ram},
url = {https://ieeexplore.ieee.org/abstract/document/8378796},
year = {2018},
date = {2018-01-01},
urldate = {2018-01-01},
booktitle = {2018 IEEE Radar Conference (RadarConf18)},
pages = {1543–1548},
organization = {IEEE},
abstract = {The detection and identification of humans and concealed objects by through wall radars is affected by wall propagation effects such as attenuation and multipath. Several works, in the past, have provided solutions for mitigating wall effects based on either prior information of the wall parameters or signal processing solutions for separating wall interference from the direct signal from the target to the radar. In this paper, we propose a machine learning based method-denoising autoencoders-to mitigate wall interference effects and for reconstructing an image resembling the ground truth in free space conditions. This method relies on training the algorithm to denoise corrupted through-wall radar images into clean line-of-sight images. We have demonstrated the effectiveness of the proposed solution using simulated narrowband Doppler-Azimuth images in free space and through-wall conditions. We simulated the propagation through diverse wall conditions using stochastic finite difference time domain techniques. Next, we tested the algorithm on measured frontal (Azimuth-Elevation) images obtained from Walabot — a wideband, low power, radar with a planar antenna array. Both the measurement and simulation results showed a low error between the denoised reconstructed images and the clean line-of-sight images.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
The detection and identification of humans and concealed objects by through wall radars is affected by wall propagation effects such as attenuation and multipath. Several works, in the past, have provided solutions for mitigating wall effects based on either prior information of the wall parameters or signal processing solutions for separating wall interference from the direct signal from the target to the radar. In this paper, we propose a machine learning based method-denoising autoencoders-to mitigate wall interference effects and for reconstructing an image resembling the ground truth in free space conditions. This method relies on training the algorithm to denoise corrupted through-wall radar images into clean line-of-sight images. We have demonstrated the effectiveness of the proposed solution using simulated narrowband Doppler-Azimuth images in free space and through-wall conditions. We simulated the propagation through diverse wall conditions using stochastic finite difference time domain techniques. Next, we tested the algorithm on measured frontal (Azimuth-Elevation) images obtained from Walabot — a wideband, low power, radar with a planar antenna array. Both the measurement and simulation results showed a low error between the denoised reconstructed images and the clean line-of-sight images.
Singh, Akash Deep; Ram, Shobha Sundar; Vishwakarma, Shelly
Simulation of the radar cross-section of dynamic human motions using virtual reality data and ray tracing Proceedings Article
In: 2018 IEEE Radar Conference (RadarConf18), pp. 1555–1560, IEEE 2018.
Abstract | Links | BibTeX | Tags:
@inproceedings{singh2018simulation,
title = {Simulation of the radar cross-section of dynamic human motions using virtual reality data and ray tracing},
author = {Akash Deep Singh and Shobha Sundar Ram and Shelly Vishwakarma},
url = {https://ieeexplore.ieee.org/abstract/document/8378798},
year = {2018},
date = {2018-01-01},
urldate = {2018-01-01},
booktitle = {2018 IEEE Radar Conference (RadarConf18)},
pages = {1555–1560},
organization = {IEEE},
abstract = {Radar returns from dynamic human motions are usually modeled using primitive based techniques. While the method is computationally simple and reasonably accurate in generating micro-Doppler signatures of humans, it is unreliable for predicting the radar cross-section (RCS) of the human especially at high frequencies. On the other hand, the shooting and bouncing ray method is effective for accurately measuring the RCS of humans. However, it has been carried out for only a single aspect or posture of the human. In this work, we present a method to simulate the radar cross-section of dynamic human motions across multiple postures by combining virtual reality data with shooting and bouncing ray (SBR) techniques. We convert each frame of human motion capture data to a poly-mesh structure of a human body and then incorporate the SBR technique for computing the resulting RCS. We verify the simulated results with measurement data at 24GHz.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Radar returns from dynamic human motions are usually modeled using primitive based techniques. While the method is computationally simple and reasonably accurate in generating micro-Doppler signatures of humans, it is unreliable for predicting the radar cross-section (RCS) of the human especially at high frequencies. On the other hand, the shooting and bouncing ray method is effective for accurately measuring the RCS of humans. However, it has been carried out for only a single aspect or posture of the human. In this work, we present a method to simulate the radar cross-section of dynamic human motions across multiple postures by combining virtual reality data with shooting and bouncing ray (SBR) techniques. We convert each frame of human motion capture data to a poly-mesh structure of a human body and then incorporate the SBR technique for computing the resulting RCS. We verify the simulated results with measurement data at 24GHz.
2017
Vishwakarma, Shelly; Ram, Shobha Sundar
Detection of multiple movers based on single channel source separation of their micro-Dopplers Journal Article
In: IEEE Transactions on Aerospace and Electronic Systems, vol. 54, no. 1, pp. 159–169, 2017.
Abstract | Links | BibTeX | Tags:
@article{vishwakarma2017detection,
title = {Detection of multiple movers based on single channel source separation of their micro-Dopplers},
author = {Shelly Vishwakarma and Shobha Sundar Ram},
url = {https://ieeexplore.ieee.org/abstract/document/8010447},
year = {2017},
date = {2017-01-01},
urldate = {2017-01-01},
journal = {IEEE Transactions on Aerospace and Electronic Systems},
volume = {54},
number = {1},
pages = {159–169},
publisher = {IEEE},
abstract = {Studies have demonstrated the usefulness of micro-Doppler signatures for classifying dynamic radar targets such as humans, helicopters, and wind turbines. However, these classification works are based on the assumption that the propagation channel consists of only a single moving target. When multiple targets move simultaneously in the channel, the micro-Dopplers, in their radar backscatter, superimpose thereby distorting the signatures. In this paper, we propose a method to detect multiple targets that move simultaneously in the propagation channel. We first model the micro-Doppler radar signatures of different movers using dictionary learning techniques. Then, we use a sparse coding algorithm to separate the aggregate radar backscatter signal from multiple targets into their individual components. We demonstrate that the disaggregated signals are useful for accurately detecting multiple targets.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Studies have demonstrated the usefulness of micro-Doppler signatures for classifying dynamic radar targets such as humans, helicopters, and wind turbines. However, these classification works are based on the assumption that the propagation channel consists of only a single moving target. When multiple targets move simultaneously in the channel, the micro-Dopplers, in their radar backscatter, superimpose thereby distorting the signatures. In this paper, we propose a method to detect multiple targets that move simultaneously in the propagation channel. We first model the micro-Doppler radar signatures of different movers using dictionary learning techniques. Then, we use a sparse coding algorithm to separate the aggregate radar backscatter signal from multiple targets into their individual components. We demonstrate that the disaggregated signals are useful for accurately detecting multiple targets.
Singh, Akash Deep; Vishwakarma, Shelly; Ram, Shobha Sundar
Co-channel interference between WiFi and through-wall micro-Doppler radar Proceedings Article
In: 2017 IEEE Radar Conference (RadarConf), pp. 1297–1302, IEEE 2017.
Abstract | Links | BibTeX | Tags:
@inproceedings{singh2017co,
title = {Co-channel interference between WiFi and through-wall micro-Doppler radar},
author = {Akash Deep Singh and Shelly Vishwakarma and Shobha Sundar Ram},
url = {https://ieeexplore.ieee.org/abstract/document/7944405},
year = {2017},
date = {2017-01-01},
urldate = {2017-01-01},
booktitle = {2017 IEEE Radar Conference (RadarConf)},
pages = {1297–1302},
organization = {IEEE},
abstract = {Narrowband through-wall radars have been researched for detecting and classifying indoor movers on the basis of their micro-Doppler signatures. These radars usually operate in the unlicensed 2.4GHz ISM band and are therefore susceptible to interference from WiFi networks operating with the IEEE 802.11g protocol. In this work, we show, through experiments, how the radar degrades the WiFi throughput by lowering the signal to noise and interference ratio at the WiFi receiver. Similarly, WiFi interference causes deterioration in the radar performance by increasing the probability of false alarms.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Narrowband through-wall radars have been researched for detecting and classifying indoor movers on the basis of their micro-Doppler signatures. These radars usually operate in the unlicensed 2.4GHz ISM band and are therefore susceptible to interference from WiFi networks operating with the IEEE 802.11g protocol. In this work, we show, through experiments, how the radar degrades the WiFi throughput by lowering the signal to noise and interference ratio at the WiFi receiver. Similarly, WiFi interference causes deterioration in the radar performance by increasing the probability of false alarms.
Vishwakarma, Shelly; Ram, Shobha Sundar
Dictionary learning for classification of indoor micro-Doppler signatures across multiple carriers Proceedings Article
In: 2017 IEEE Radar Conference (RadarConf), pp. 0992–0997, IEEE 2017.
Abstract | Links | BibTeX | Tags:
@inproceedings{vishwakarma2017dictionary,
title = {Dictionary learning for classification of indoor micro-Doppler signatures across multiple carriers},
author = {Shelly Vishwakarma and Shobha Sundar Ram},
url = {https://ieeexplore.ieee.org/abstract/document/7944348},
year = {2017},
date = {2017-01-01},
urldate = {2017-01-01},
booktitle = {2017 IEEE Radar Conference (RadarConf)},
pages = {0992–0997},
organization = {IEEE},
abstract = {Micro-Doppler signatures of dynamic targets such as humans, animals and vehicles are very effective feature vectors for classification based on machine learning algorithms. In the existing works, the test data have been measured in nearly identical operating conditions to the training data that were gathered for the classifiers. However, this assumption may be violated in real life scenarios. In this work, we demonstrate that classification based on sparsity based dictionary learning can overcome this limitation. Here, we learn unique target class dictionaries from micro-Dopplers gathered at multiple carriers. Then we test the classifier using data gathered at another carrier (distinct from those used for training).We test the performance of the classification algorithm for both simulation and measurement data. Our results show a classification accuracy of 99% and 89% for simulated and measurement data respectively.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Micro-Doppler signatures of dynamic targets such as humans, animals and vehicles are very effective feature vectors for classification based on machine learning algorithms. In the existing works, the test data have been measured in nearly identical operating conditions to the training data that were gathered for the classifiers. However, this assumption may be violated in real life scenarios. In this work, we demonstrate that classification based on sparsity based dictionary learning can overcome this limitation. Here, we learn unique target class dictionaries from micro-Dopplers gathered at multiple carriers. Then we test the classifier using data gathered at another carrier (distinct from those used for training).We test the performance of the classification algorithm for both simulation and measurement data. Our results show a classification accuracy of 99% and 89% for simulated and measurement data respectively.
2016
Vishwakarma, Shelly; Ram, Shobha Sundar
Classification of multiple targets based on disaggregation of micro-doppler signatures Proceedings Article
In: 2016 Asia-Pacific Microwave Conference (APMC), pp. 1–4, IEEE 2016.
Abstract | Links | BibTeX | Tags:
@inproceedings{vishwakarma2016classification,
title = {Classification of multiple targets based on disaggregation of micro-doppler signatures},
author = {Shelly Vishwakarma and Shobha Sundar Ram},
url = {https://ieeexplore.ieee.org/abstract/document/7931360},
year = {2016},
date = {2016-01-01},
urldate = {2016-01-01},
booktitle = {2016 Asia-Pacific Microwave Conference (APMC)},
pages = {1–4},
organization = {IEEE},
abstract = {Micro-Doppler signatures of dynamic indoor targets (such as humans and fans) serve as a useful tool for classification. However, all the current classification methods are limited by the assumption that only a single target is present in the channel. In this work, we propose a method to classify multiple targets that are simultaneously present in the channel on the basis of a single channel source separation technique. We apply sparse coding based dictionary learning (DL) algorithms for disaggregating micro-Doppler returns from multiple targets into its constituent signals. The classification is subsequently carried out on the disaggregated signals. We have tested the performance of the proposed algorithm on simulated human and fan data.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Micro-Doppler signatures of dynamic indoor targets (such as humans and fans) serve as a useful tool for classification. However, all the current classification methods are limited by the assumption that only a single target is present in the channel. In this work, we propose a method to classify multiple targets that are simultaneously present in the channel on the basis of a single channel source separation technique. We apply sparse coding based dictionary learning (DL) algorithms for disaggregating micro-Doppler returns from multiple targets into its constituent signals. The classification is subsequently carried out on the disaggregated signals. We have tested the performance of the proposed algorithm on simulated human and fan data.