Publications

@article{salami2022tesla,
  title={Tesla-rapture: A lightweight gesture recognition system from mmwave radar sparse point clouds},
  author={Salami, Dariush and Hasibi, Ramin and Palipana, Sameera and Popovski, Petar and Michoel, Tom and Sigg, Stephan},
  journal={IEEE Transactions on Mobile Computing},
  year={2022},
  publisher={IEEE},
  group = {ambience},
  project = {radiosense}
}

@article{santoboni2022wireless,
  title={Wireless LAN sensing with smart antennas},
  author={Santoboni, Marco and Bersan, Riccardo and Savazzi, Stefano and Zecchin, Alberto and Piazza, Vittorio Rampa Daniele},
  journal={arXiv preprint arXiv:2205.00973},
      project = {radiosense},
  year={2022}
}


  

@article{zhang2022practical,
  title={Practical Issues and Challenges in CSI-based Integrated Sensing and Communication},
  author={Zhang, Daqing and Wu, Dan and Niu, Kai and Wang, Xuanzhi and Zhang, Fusang and Yao, Jian and Jiang, Dajie and Qin, Fei},
  journal={arXiv preprint arXiv:2204.03535},
   project = {radiosense},
  year={2022}
}

@article{niu2022rethinking,
  title={Rethinking Doppler Effect for Accurate Velocity Estimation with Commodity WiFi Devices},
  author={Niu, Kai and Wang, Xuanzhi and Zhang, Fusang and Zheng, Rong and Yao, Zhiyun and Zhang, Daqing},
  journal={IEEE Journal on Selected Areas in Communications},
  year={2022},
   project = {radiosense},
  publisher={IEEE}
}

@incollection{zhang2022wifi,
  title={WiFi CSI-based vital signs monitoring},
  author={Zhang, Daqing and Zeng, Youwei and Zhang, Fusang and Xiong, Jie},
  booktitle={Contactless Vital Signs Monitoring},
  pages={231--255},
  year={2022},
project = {radiosense},
  publisher={Elsevier}
}

  

@InProceedings{Salami_2021_MLSP, 
author = {Dariush Salami and Stephan Sigg}, 
booktitle = {IEEE 31st International Workshop on Machine Learning for Signal Processing (MLSP)}, 
title = {Zero-shot Motion Pattern Recognition from 4D Point Clouds},
year = {2021},
project ={radiosense,windmill},
group={ambience}
}

@inproceedings{Sigg2020Camouflage,
  title={Camouflage Learning},
  author={Stephan Sigg and Le Ngu Nguyen and Jing Ma},
  booktitle={The 19th International Conference on Pervasive Computing and Communications (PerCom 2021), adjunct},
  year={2021},
  abstract={Federated learning has been proposed as a concept for distributed machine learning which enforces privacy by avoiding sharing private data with a coordinator or distributed nodes. Instead of gathering datasets to a central server for model training in traditional machine learning, in federated learning, model updates are computed locally at distributed devices and merged at a coordinator. However, information on local data might be leaked through the model updates. We propose Camouflage learning, a distributed machine learning scheme that distributes both the data and the model. Neither the distributed devices nor the coordinator is at any point in time in possession of the complete model. Furthermore, data and model are obfuscated during distributed model inference and distributed model training. Camouflage learning can be implemented with various Machine learning schemes.
},
  group = {ambience},
  project = {radiosense, abacus}
}

Federated learning has been proposed as a concept for distributed machine learning which enforces privacy by avoiding sharing private data with a coordinator or distributed nodes. Instead of gathering datasets to a central server for model training in traditional machine learning, in federated learning, model updates are computed locally at distributed devices and merged at a coordinator. However, information on local data might be leaked through the model updates. We propose Camouflage learning, a distributed machine learning scheme that distributes both the data and the model. Neither the distributed devices nor the coordinator is at any point in time in possession of the complete model. Furthermore, data and model are obfuscated during distributed model inference and distributed model training. Camouflage learning can be implemented with various Machine learning schemes.

@inproceedings{Manila2020BatteryLess,
  title={Towards battery-less RF sensing},
  author={Manila Kodali and Le Ngu Nguyen and Stephan Sigg},
  booktitle={The 19th International Conference on Pervasive Computing and Communications (PerCom 2021), WiP},
  year={2021},
  abstract={Recent work has demonstrated the use of the radio interface as a sensing modality for gestures, activities and situational perception. The field generally moves towards larger bandwidths, multiple antennas, and higher, mmWave frequency domains, which allow for the recognition of minute movements. We envision another set of applications for RF sensing: battery-less autonomous sensing devices. In this work, we investigate transceiver-less passive RF-sensors which are excited by the fluctuation of the received power over the wireless channel. In particular, we demonstrate the use of battery-less RF-sensing for applications of on-body gesture recognition integrated into smart garment, as well as the integration of such sensing capabilities into smart surfaces.
},
  group = {ambience},
  project = {radiosense,abacus}
}

Recent work has demonstrated the use of the radio interface as a sensing modality for gestures, activities and situational perception. The field generally moves towards larger bandwidths, multiple antennas, and higher, mmWave frequency domains, which allow for the recognition of minute movements. We envision another set of applications for RF sensing: battery-less autonomous sensing devices. In this work, we investigate transceiver-less passive RF-sensors which are excited by the fluctuation of the received power over the wireless channel. In particular, we demonstrate the use of battery-less RF-sensing for applications of on-body gesture recognition integrated into smart garment, as well as the integration of such sensing capabilities into smart surfaces.

@article{Sameera_2021_IMWUT,
author={Sameera Palipana and Dariush Salami and Luis Leiva and Stephan Sigg},
journal={Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT)},
title={Pantomime: Mid-Air Gesture Recognition with Sparse Millimeter-Wave Radar Point Clouds},
year={2021},
abstract={We introduce Pantomime, a novel mid-air gesture recognition system exploiting spatio-temporal properties of millimeter-wave radio frequency (RF) signals. Pantomime is positioned in a unique region of the RF landscape: mid-resolution mid-range high-frequency sensing, which makes it ideal for motion gesture interaction. We configure a commercial frequency-modulated continuous-wave radar device to promote spatial information over temporal resolution by means of sparse 3D point clouds, and contribute a deep learning architecture that directly consumes the point cloud, enabling real-time performance with low computational demands. Pantomime achieves 95\% accuracy and 99\% AUC in a challenging set of 21 gestures articulated by 45 participants in two indoor environments, outperforming four state-of-the-art 3D point cloud recognizers. We also analyze the effect of environment, articulation speed, angle, and distance to the sensor. We conclude that Pantomime is resilient to various input conditions and that it may enable novel applications in industrial, vehicular, and smart home scenarios.
},
issue_date = {March 2021},
publisher = {ACM New York, NY, USA},
volume = {5},
number = {1},
pages = {1-27},
group = {ambience},
project = {radiosense,windmill}
}

We introduce Pantomime, a novel mid-air gesture recognition system exploiting spatio-temporal properties of millimeter-wave radio frequency (RF) signals. Pantomime is positioned in a unique region of the RF landscape: mid-resolution mid-range high-frequency sensing, which makes it ideal for motion gesture interaction. We configure a commercial frequency-modulated continuous-wave radar device to promote spatial information over temporal resolution by means of sparse 3D point clouds, and contribute a deep learning architecture that directly consumes the point cloud, enabling real-time performance with low computational demands. Pantomime achieves 95% accuracy and 99% AUC in a challenging set of 21 gestures articulated by 45 participants in two indoor environments, outperforming four state-of-the-art 3D point cloud recognizers. We also analyze the effect of environment, articulation speed, angle, and distance to the sensor. We conclude that Pantomime is resilient to various input conditions and that it may enable novel applications in industrial, vehicular, and smart home scenarios.

@inproceedings{rampa2021automatic,
  title={Automatic Calibration of Anchor Nodes in Device-Free Radio Localization and Motion Tracking Scenarios},
  author={Rampa, Vittorio and Savazzi, Stefano and D’Amico, Michele},
  booktitle={2021 IEEE-APS Topical Conference on Antennas and Propagation in Wireless Communications (APWC)},
  pages={137--142},
  year={2021},
    project = {radiosense},
  organization={IEEE}
}

@article{chen2021constructing,
  title={Constructing Floor Plan through Smoke Using Ultra Wideband Radar},
  author={Chen, Weiyan and Zhang, Fusang and Gu, Tao and Zhou, Kexing and Huo, Zixuan and Zhang, Daqing},
  journal={Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies},
  volume={5},
  number={4},
  pages={1--29},
  year={2021},
    project = {radiosense},
  publisher={ACM New York, NY, USA}
}
  
  
  

@inproceedings{savazzi2021framework,
  title={A framework for energy and carbon footprint analysis of distributed and federated edge learning},
  author={Savazzi, Stefano and Kianoush, Sanaz and Rampa, Vittorio and Bennis, Mehdi},
  booktitle={2021 IEEE 32nd Annual International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC)},
  pages={1564--1569},
  year={2021},
   project = {radiosense},
  organization={IEEE}
}

@article{rampa2021electromagnetic,
  title={Electromagnetic models for passive detection and localization of multiple bodies},
  author={Rampa, Vittorio and Gentili, Gian Guido and Savazzi, Stefano and D'Amico, Michele},
  journal={arXiv preprint arXiv:2104.07354},
   project = {radiosense},
  year={2021}
}

@article{savazzi2021opportunities,
  title={Opportunities of federated learning in connected, cooperative, and automated industrial systems},
  author={Savazzi, Stefano and Nicoli, Monica and Bennis, Mehdi and Kianoush, Sanaz and Barbieri, Luca},
  journal={IEEE Communications Magazine},
  volume={59},
  number={2},
  pages={16--21},
  year={2021},
   project = {radiosense},
  publisher={IEEE}
}

@article{rampa2021model,
  title={EM model-based device-free localization of multiple bodies},
  author={Rampa, Vittorio and Nicoli, Monica and Manno, Chiara and Savazzi, Stefano},
  journal={Sensors},
  volume={21},
  number={5},
  pages={1728},
  year={2021},
   project = {radiosense},
  publisher={Multidisciplinary Digital Publishing Institute}
}

@article{zhang2021exploring,
  title={Exploring LoRa for Sensing},
  author={Zhang, Fusang and Chang, Zhaoxin and Xiong, Jie and Zhang, Daqing},
  journal={GetMobile: Mobile Computing and Communications},
  volume={25},
  number={2},
  pages={33--37},
  year={2021},
   project = {radiosense},
  publisher={ACM New York, NY, USA}
}

@PhDThesis{MuneebaThesis2020,
	author = "Muneeba Raja",
	title = "Toward Complex 3D Movement Detection to Analyze Human Behavior via Radio-Frequency Signals",
	school = "Aalto University",
	year = "2020",
	month = "September",
	isbn = "978-952-60-3988-6",
    url_Paper ={https://aaltodoc.aalto.fi/handle/123456789/46311},
    abstract = {A driver's attention, parallel actions, and emotions directly influence driving behavior. Any secondary task, be it cognitive, visual, or manual, that diverts driver focus from the primary task of driving is a source of distraction. Longer response time, inability to scan the road, and missing visual cues can all lead to car crashes with serious consequences. Current research focuses on detecting distraction by means of vehicle-mounted video cameras or wearable sensors for tracking eye movements and head rotation. Facial expressions, speech, and physiological signals are also among the widely used indicators for detecting distraction. These approaches are accurate, fast, and reliable but come with a high installation cost, requirements related to lighting conditions, privacy intrusions, and energy consumption.
 
Over the past decade, the use of radio signals has been investigated as a possible solution for the aforementioned limitations of today's technologies. Changes in radio-signal patterns caused by movements of the human body can be analyzed and thereby used in detecting humans' gestures and activities. Human behavior and emotions, in particular, are less explored in this regard and are addressed mostly with reference to physiological signals.
 
The thesis exploited multiple wireless technologies (1.8~GHz, WiFi, and millimeter wave) and combinations thereof to detect complex 3D movements of a driver in a car. Upper-body movements are vital indicators of a driver's behavior in a car, and the information from these movements could be used to generate appropriate feedback, such as warnings or provision of directives for actions that would avoid jeopardizing safety. Existing wireless-system-based solutions focus primarily on either large or small movements, or they address well-defined activities. They do not consider discriminating large movements from small ones, let alone their directions, within a single system. These limitations underscore the requirement to address complex natural-behavior situations precisely such as that in a car, which demands not only isolating particular movements but also classifying and predicting them.
 
The research to reach the attendant goals exploited physical properties of RF signals, several hardware-software combinations, and building of algorithms to process and detect body movements -- from the simple to the complex. Additionally, distinctive feature sets were addressed for machine-learning techniques to find patterns in data and predict states accordingly. The systems were evaluated by performing extensive real-world studies.},
group = {ambience}, 
project = {radiosense}}

A driver's attention, parallel actions, and emotions directly influence driving behavior. Any secondary task, be it cognitive, visual, or manual, that diverts driver focus from the primary task of driving is a source of distraction. Longer response time, inability to scan the road, and missing visual cues can all lead to car crashes with serious consequences. Current research focuses on detecting distraction by means of vehicle-mounted video cameras or wearable sensors for tracking eye movements and head rotation. Facial expressions, speech, and physiological signals are also among the widely used indicators for detecting distraction. These approaches are accurate, fast, and reliable but come with a high installation cost, requirements related to lighting conditions, privacy intrusions, and energy consumption. Over the past decade, the use of radio signals has been investigated as a possible solution for the aforementioned limitations of today's technologies. Changes in radio-signal patterns caused by movements of the human body can be analyzed and thereby used in detecting humans' gestures and activities. Human behavior and emotions, in particular, are less explored in this regard and are addressed mostly with reference to physiological signals. The thesis exploited multiple wireless technologies (1.8 GHz, WiFi, and millimeter wave) and combinations thereof to detect complex 3D movements of a driver in a car. Upper-body movements are vital indicators of a driver's behavior in a car, and the information from these movements could be used to generate appropriate feedback, such as warnings or provision of directives for actions that would avoid jeopardizing safety. Existing wireless-system-based solutions focus primarily on either large or small movements, or they address well-defined activities. They do not consider discriminating large movements from small ones, let alone their directions, within a single system. These limitations underscore the requirement to address complex natural-behavior situations precisely such as that in a car, which demands not only isolating particular movements but also classifying and predicting them. The research to reach the attendant goals exploited physical properties of RF signals, several hardware-software combinations, and building of algorithms to process and detect body movements – from the simple to the complex. Additionally, distinctive feature sets were addressed for machine-learning techniques to find patterns in data and predict states accordingly. The systems were evaluated by performing extensive real-world studies.

@article{10.1145/3397326,
author = {Zhang, Fusang and Chang, Zhaoxin and Niu, Kai and Xiong, Jie and Jin, Beihong and Lv, Qin and Zhang, Daqing},
title = {Exploring LoRa for Long-Range Through-Wall Sensing},
year = {2020},
issue_date = {June 2020},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
volume = {4},
number = {2},
journal = {Proc. ACM Interact. Mob. Wearable Ubiquitous Technol.},
month = jun,
articleno = {68},
numpages = {27},
project={radiosense}
}

@article{Sanaz_2020_IoT,
author={Sanaz Kianoush and Stefano Savazzi and Manuel Beschi and Stephan Sigg and Vittorio Rampa},
journal={IEEE Internet of Things Journal},
title={A Multisensory Edge-Cloud Platform for Opportunistic Sensing in Cobot Environments},
year={2020},
doi = {10.1109/JIOT.2020.3011809},
project={radiosense},
group = {ambience}
}

@article{savazzi2020federated,
  title={Federated learning with cooperating devices: A consensus approach for massive IoT networks},
  author={Savazzi, Stefano and Nicoli, Monica and Rampa, Vittorio},
  journal={IEEE Internet of Things Journal},
  volume={7},
  number={5},
  pages={4641--4654},
  year={2020},
  project={radiosense},
  publisher={IEEE}
}

@inproceedings{savazzi2020federated2,
  title={Federated Learning with Mutually Cooperating Devices: A Consensus Approach Towards Server-Less Model Optimization},
  author={Savazzi, Stefano and Nicoli, Monica and Rampa, Vittorio and Kianoush, Sanaz},
  booktitle={ICASSP 2020-2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  pages={3937--3941},
  year={2020},
  project={radiosense},
  organization={IEEE}
}

@inproceedings{savazzi2020Joint,
  title={A Joint Decentralized Federated Learning and Communications Framework for Industrial Networks},
  author={Stefano Savazzi and Sanaz Kianoush and Vittorio Rampa and M. Bennis},
  booktitle={IEEE International Workshop on Computer Aided Modeling and Design of Communication Links and Networks},
  year={2020},
  project={radiosense}
}

@InProceedings{Salami_2020_MLSP, 
author = {Dariush Salami and Sameera Palipana and Manila Kodali and Stephan Sigg}, 
booktitle = {IEEE 30th International Workshop on Machine Learning for Signal Processing (MLSP)}, 
title = {Motion Pattern Recognition in 4D Point Clouds},
year = {2020},
project ={radiosense,windmill},
group={ambience}
}

@article{Muneeba_2020_3D,
author={ Muneeba Raja and Zahra Vali and Sameera Palipana and David G. Michelson and Stephan Sigg },
journal={IEEE Access},
title={3D head motion detection using millimeter-wave Doppler radar},
year={2020},
doi = {10.1109/ACCESS.2020.2973957},
project={radiosense},
group = {ambience}
}

@InProceedings{Sameera_2020_Beamsteering,
  author    = {Sameera Palipana and Nicolas Malm and Stephan Sigg},
  booktitle = {18th Annual IEEE International Conference on Pervasive Computing and Communications (PerCom) },
  title     = {Beamsteering for training-free Recognition of Multiple Humans Performing Distinct Activities},
  year      = {2020},
  doi = {10.1109/PerCom45495.2020.9127374},
  abstract  = {Recognition of the context of humans plays an important role in pervasive applications such as intrusion detection, human density estimation for heating, ventilation and air-conditioning in smart buildings, as well as safety guarantee for workers during human-robot interaction. Radio vision is able to provide these sensing capabilities with low privacy intrusion. A common challenge though, for current radio sensing solutions is to distinguish simultaneous movement from multiple subjects. We present an approach that exploits multi-antenna installations, for instance, found in upcoming 5G instrumentations, to detect and extract activities from spatially scattered human targets in an ad-hoc manner in arbitrary environments and without prior training of the multi-subject detection. We perform receiver-side beamforming and beam-steering over different azimuth angles to detect human presence in those regions separately. We characterize the resultant fluctuations in the spatial streams due to human influence using a case study and make the traces publicly available. We demonstrate the potential of this approach through two applications: 1) By feeding the similarities of the resulting spatial streams into a clustering algorithm, we count the humans in a given area without prior training. (up to 6 people in a 22.4m2 area with an accuracy that significantly exceeds the related work). 2) We further demonstrate that simultaneously conducted activities and gestures can be extracted from the spatial streams through blind source separation.},
  %url_Paper = {http://ambientintelligence.aalto.fi/paper/findling_closed_eye_eog.pdf},
  project = {radiosense},
  group = {ambience}
  }

Recognition of the context of humans plays an important role in pervasive applications such as intrusion detection, human density estimation for heating, ventilation and air-conditioning in smart buildings, as well as safety guarantee for workers during human-robot interaction. Radio vision is able to provide these sensing capabilities with low privacy intrusion. A common challenge though, for current radio sensing solutions is to distinguish simultaneous movement from multiple subjects. We present an approach that exploits multi-antenna installations, for instance, found in upcoming 5G instrumentations, to detect and extract activities from spatially scattered human targets in an ad-hoc manner in arbitrary environments and without prior training of the multi-subject detection. We perform receiver-side beamforming and beam-steering over different azimuth angles to detect human presence in those regions separately. We characterize the resultant fluctuations in the spatial streams due to human influence using a case study and make the traces publicly available. We demonstrate the potential of this approach through two applications: 1) By feeding the similarities of the resulting spatial streams into a clustering algorithm, we count the humans in a given area without prior training. (up to 6 people in a 22.4m2 area with an accuracy that significantly exceeds the related work). 2) We further demonstrate that simultaneously conducted activities and gestures can be extracted from the spatial streams through blind source separation.

@article{Lin_2020_Contactless,
author={Nan Lin and Yangjie Cao and Fuchao Wang and Xinxin Lu and Bo Zhang and Zhi Liu and Stephan Sigg},
journal={IEEE Internet of Things Journal},
title={Contactless Body Movement Recognition during Sleep via WiFi Signals},
year={2020},
doi = {10.1109/JIOT.2019.2960823},
group = {ambience},
project = {radiosense}
}

@article{zhang2020your,
  title={Your Smart Speaker Can" Hear" Your Heartbeat!},
  author={Zhang, Fusang and Wang, Zhi and Jin, Beihong and Xiong, Jie and Zhang, Daqing},
  journal={Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies},
  volume={4},
  number={4},
  pages={1--24},
  year={2020},
    project = {radiosense},
  publisher={ACM New York, NY, USA}
}

@article{zeng2020boosting,
  title={Boosting WiFi sensing performance via CSI ratio},
  author={Zeng, Youwei and Wu, Dan and Xiong, Jie and Zhang, Daqing},
  journal={IEEE Pervasive Computing},
  volume={20},
  number={1},
  pages={62--70},
  year={2020},
  project = {radiosense},
  publisher={IEEE}
}
  
  

@article{zhang2020exploring,
  title={Exploring lora for long-range through-wall sensing},
  author={Zhang, Fusang and Chang, Zhaoxin and Niu, Kai and Xiong, Jie and Jin, Beihong and Lv, Qin and Zhang, Daqing},
  journal={Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies},
  volume={4},
  number={2},
  pages={1--27},
  year={2020},
   project = {radiosense},
  publisher={ACM New York, NY, USA}
}

@article{savazzi2020processing,
  title={Processing of body-induced thermal signatures for physical distancing and temperature screening},
  author={Savazzi, Stefano and Rampa, Vittorio and Costa, Leonardo and Kianoush, Sanaz and Tolochenko, Denis},
  journal={IEEE sensors Journal},
  volume={21},
  number={13},
  pages={14168--14179},
  year={2020},
project = {radiosense},
  publisher={IEEE}
}

@article{wu2020fingerdraw,
  title={FingerDraw: Sub-wavelength Level Finger Motion Tracking with WiFi Signals},
  author={Dan Wu and Guiyang Gao and Youwei Zeng and Jinyi Liu and Leye Wang and Tao Gu and Daqing Zhang},
  journal={Proc. ACM Interact. Mob. Wearable Ubiquitous Technol},
  volume={1},
  number={1},
  year={2020},
  project = {radiosense}
}

@article{chen2020robust,
  title={Robust Dynamic Hand Gesture Interaction using LTE Terminals},
  author={Weiyan Chen and Kai Niu and Deng Zhao and Rong Zheng and Dan Wu and Wei Wang and Leye Wang and Daqing Zhang},
  booktitle={Proceedings of the ACM/IEEE Conference on Informtion Processing in Sensor Networks (IPSN)},
  year={2020},
  project = {radiosense}
}

@article{Qiyue_2019_DCGAN,
author={Qiyue Li and Heng Qu and Zhi Liu and Nana Zhou and Wei Sun and Stephan Sigg and Jie Li},
journal={IEEE Transactions on Emerging Topics in Computational Intelligence},
title={AF-DCGAN: Amplitude Feature Deep Convolutional GAN for Fingerprint Construction in Indoor Localization Systems},
year={2019},
abstract = {With widely deployed WiFi network and the uniqueness feature (fingerprint) of wireless channel information, fingerprinting based WiFi positioning is currently the mainstream indoor positioning method, in which fingerprint database construction is crucial. 
However, for accuracy, this approach requires enough data to be sampled at many reference points, which consumes excessive efforts and time. 
In this paper, we collect Channel State Information (CSI) data at reference points by the method of device-free localization, then we convert collected CSI data into amplitude feature maps and extend the fingerprint
database using the proposed Amplitude-Feature Deep Convolutional Generative Adversarial Network (AF-DCGAN) model. 
The use of AF-DCGAN accelerates convergence during the training phase, and substantially increases the diversity of the CSI amplitude feature map. The extended fingerprint database both reduces the human effort involved in fingerprint database construction and the accuracy of an indoor localization system, as demonstrated in the experiments.
},
doi =  {10.1109/TETCI.2019.2948058},
project = {radiosense},
group = {ambience}
}

With widely deployed WiFi network and the uniqueness feature (fingerprint) of wireless channel information, fingerprinting based WiFi positioning is currently the mainstream indoor positioning method, in which fingerprint database construction is crucial. However, for accuracy, this approach requires enough data to be sampled at many reference points, which consumes excessive efforts and time. In this paper, we collect Channel State Information (CSI) data at reference points by the method of device-free localization, then we convert collected CSI data into amplitude feature maps and extend the fingerprint database using the proposed Amplitude-Feature Deep Convolutional Generative Adversarial Network (AF-DCGAN) model. The use of AF-DCGAN accelerates convergence during the training phase, and substantially increases the diversity of the CSI amplitude feature map. The extended fingerprint database both reduces the human effort involved in fingerprint database construction and the accuracy of an indoor localization system, as demonstrated in the experiments.

@article{Sameera_2019_access,
author={Sameera Palipana and Stephan Sigg},
journal={IEEE Access},
title={Extracting Human Context through Receiver-end Beamforming},
year={2019},
abstract = {Device-free passive sensing of the human targets using wireless signals have acquired much attention in the recent past because of its importance in many applications including security, heating, ventilation
and air conditioning (HVACs), activity recognition, and elderly care. In this paper, we use receiver-side beamforming to isolate the array response of a human target when the line of sight array response is several
magnitudes stronger than the human response. The solution is implemented in a 5G testbed using a software-defined radio (SDR) platform. As beamforming with SDRs faces the challenge to train the beamformer to
different azimuth angles, we present an algorithm to generate the steering vectors for all azimuth angles from a few training directions amidst imprecise prior information on the training steering vectors. We extract the
direction of arrival (DoA) from the array response of the human target, and conducting experiments in a semi-anechoic chamber, we detect the DoAs of up to four stationary human targets and track the DoA of up
to two walking persons simultaneously.
},
  project = {radiosense},
group = {ambience}
}

Device-free passive sensing of the human targets using wireless signals have acquired much attention in the recent past because of its importance in many applications including security, heating, ventilation and air conditioning (HVACs), activity recognition, and elderly care. In this paper, we use receiver-side beamforming to isolate the array response of a human target when the line of sight array response is several magnitudes stronger than the human response. The solution is implemented in a 5G testbed using a software-defined radio (SDR) platform. As beamforming with SDRs faces the challenge to train the beamformer to different azimuth angles, we present an algorithm to generate the steering vectors for all azimuth angles from a few training directions amidst imprecise prior information on the training steering vectors. We extract the direction of arrival (DoA) from the array response of the human target, and conducting experiments in a semi-anechoic chamber, we detect the DoAs of up to four stationary human targets and track the DoA of up to two walking persons simultaneously.

@InProceedings{Palipana_2019_buildsys,
author={Sameera Palipana and Stephan Sigg},
booktitle={Conference on Systems for Energy-Efficient Buildings, Cities and Transportation (Buildsys) (adjunct)},
title={Receiver-Side Beamforming to Isolate Channel Perturbations from a Human Target in a Device-Free Setting},
year={2019},
abstract={We present an approach to isolate the angular response of a human on a receiver-side beamformer when the line of sight is sevaral magnitudes stronger than the human response. 
The solution is implemented in a 5G testbed using a software-defined radio (SDR) platform. 
Beamforming with SDRs faces the challenge to train the beamformer to different azimuth angles. 
We present an algorithm to generate the steering vectors from a few training points amidst imprecise prior information. 
In particular, we assign asimuth angles to steering vectors converted from phase rotations of signals transmitted from reference directions. 
Furthermore, we detect a human and estimate the direction from strong signal perturbations towards that angle. 
Experiments for a person performing in-place activities in a semi-anechoic chamber show a detection accuracy of 100% and a maximum median direction of arrival error of 40 degree.},
  project = {radiosense},
group = {ambience}
}

We present an approach to isolate the angular response of a human on a receiver-side beamformer when the line of sight is sevaral magnitudes stronger than the human response. The solution is implemented in a 5G testbed using a software-defined radio (SDR) platform. Beamforming with SDRs faces the challenge to train the beamformer to different azimuth angles. We present an algorithm to generate the steering vectors from a few training points amidst imprecise prior information. In particular, we assign asimuth angles to steering vectors converted from phase rotations of signals transmitted from reference directions. Furthermore, we detect a human and estimate the direction from strong signal perturbations towards that angle. Experiments for a person performing in-place activities in a semi-anechoic chamber show a detection accuracy of 100% and a maximum median direction of arrival error of 40 degree.

@InProceedings{Sigg_2019_miel,
author={Stephan Sigg and Sameera Palipana and Stefano Savazzi and Sanaz Kianoush},
booktitle={International Conference on Business Process Management (adjunct)},
title={Capturing human-machine interaction events from radio sensors in Industry 4.0 environments},
year={2019}, 
abstract={In manufacturing environments, human workers interact with increasingly autonomous machinery. 
To ensure workspace safety and production efficiency during human-robot cooperation, continuous and accurate tracking and perception of workers activities is required. 
The RadioSense project intends to move forward the state-of-the-art in advanced sensing and perception for next generation manufacturing workspace. 
In this paper, we describe our ongoing efforts towards multi-subject recognition cases with multiple persons conducting several simultaneous activities. 
Perturbations induced by moving bodies/objects on the electro-magnetic wavefield can be processed for environmental perception. 
In particular, we will adopt next generation (5G) high-frequency technologies as well as distributed massive MIMO systems.
},
  project = {radiosense},
group = {ambience}}

In manufacturing environments, human workers interact with increasingly autonomous machinery. To ensure workspace safety and production efficiency during human-robot cooperation, continuous and accurate tracking and perception of workers activities is required. The RadioSense project intends to move forward the state-of-the-art in advanced sensing and perception for next generation manufacturing workspace. In this paper, we describe our ongoing efforts towards multi-subject recognition cases with multiple persons conducting several simultaneous activities. Perturbations induced by moving bodies/objects on the electro-magnetic wavefield can be processed for environmental perception. In particular, we will adopt next generation (5G) high-frequency technologies as well as distributed massive MIMO systems.

@article{Savazzi_2019_transformative,
author={Stefano Savazzi and Stephan Sigg and Federico Vicentini and Sanaz Kianoush and Rainhard Findling},
journal={IEEE Computer, SI on on Transformative Computing and Communication},
title={On the use of stray wireless signals for sensing: a look beyond 5G for the next generation industry},
year={2019},
  number = {7},
  pages = {25-36},
  volume = {52},
  doi = {10.1109/MC.2019.2913626},
abstract = {Transformative techniques to capture and process wireless stray radiation originated from different radio sources are gaining increasing attention. They can be applied to human sensing, behavior recognition, localization and mapping. The omnipresent radio-frequency (RF) stray radiation of wireless devices (WiFi, Cellular or any Personal/Body Area Network) encodes a 3D view of all objects traversed by its propagation. A trained machine learning model is then applied to features extracted in real-time from radio signals to isolate body-induced footprints or environmental alterations. The technology can augment and transform existing radio-devices into ubiquitously distributed sensors that simultaneously act as wireless transmitters and receivers (e.g. fast time-multiplexed). Thereby, 5G-empowered tiny device networks transform into a dense web of RF-imaging links that extract a view of an environment, for instance, to monitor manufacturing processes in next generation industrial set-ups (Industry 4.0, I4.0). This article highlights emerging transformative computing tools for radio sensing, promotes key technology enablers in 5G communication and reports deployment experiences.},
  project = {radiosense},
group = {ambience}}

Transformative techniques to capture and process wireless stray radiation originated from different radio sources are gaining increasing attention. They can be applied to human sensing, behavior recognition, localization and mapping. The omnipresent radio-frequency (RF) stray radiation of wireless devices (WiFi, Cellular or any Personal/Body Area Network) encodes a 3D view of all objects traversed by its propagation. A trained machine learning model is then applied to features extracted in real-time from radio signals to isolate body-induced footprints or environmental alterations. The technology can augment and transform existing radio-devices into ubiquitously distributed sensors that simultaneously act as wireless transmitters and receivers (e.g. fast time-multiplexed). Thereby, 5G-empowered tiny device networks transform into a dense web of RF-imaging links that extract a view of an environment, for instance, to monitor manufacturing processes in next generation industrial set-ups (Industry 4.0, I4.0). This article highlights emerging transformative computing tools for radio sensing, promotes key technology enablers in 5G communication and reports deployment experiences.

@article{Raja_2019_antenna,
author={Muneeba Raja and Aidan Hughes and Yixuan Xu and Parham zarei and David G. Michelson and Stephan Sigg},
journal={IEEE Antennas and Wireless Propagation letters},
title={Wireless Multi-frequency Feature Set to Simplify Human 3D Pose Estimation},
year={2019},
volume={18},
number={5},
pages={876-880},
doi = {10.1109/LAWP.2019.2904580},
abstract = {We present a multifrequency feature set to detect driver's three-dimensional (3-D) head and torso movements from fluctuations in the radio frequency channel due to body movements. Current features used for movement detection are based on the time-of-flight, received signal strength, and channel state information and come with the limitations of coarse tracking, sensitivity toward multipath effects, and handling corrupted phase data, respectively. There is no standalone feature set that accurately detects small and large movements and determines the direction in 3-D space. We resolve this problem by using two radio signals at widely separated frequencies in a monostatic configuration. By combining information about displacement, velocity, and direction of movements derived from the Doppler effect at each frequency, we expand the number of existing features. We separate pitch, roll, and yaw movements of head from torso and arm. The extracted feature set is used to train a K-Nearest Neighbor classification algorithm, which could provide behavioral awareness to cars while being less invasive as compared to camera-based systems. The training results on data from four participants reveal that the classification accuracy is 77.4% at 1.8 GHz, it is 87.4% at 30 GHz, and multifrequency feature set improves the accuracy to 92%.},
  project = {radiosense},
group = {ambience}}

We present a multifrequency feature set to detect driver's three-dimensional (3-D) head and torso movements from fluctuations in the radio frequency channel due to body movements. Current features used for movement detection are based on the time-of-flight, received signal strength, and channel state information and come with the limitations of coarse tracking, sensitivity toward multipath effects, and handling corrupted phase data, respectively. There is no standalone feature set that accurately detects small and large movements and determines the direction in 3-D space. We resolve this problem by using two radio signals at widely separated frequencies in a monostatic configuration. By combining information about displacement, velocity, and direction of movements derived from the Doppler effect at each frequency, we expand the number of existing features. We separate pitch, roll, and yaw movements of head from torso and arm. The extracted feature set is used to train a K-Nearest Neighbor classification algorithm, which could provide behavioral awareness to cars while being less invasive as compared to camera-based systems. The training results on data from four participants reveal that the classification accuracy is 77.4% at 1.8 GHz, it is 87.4% at 30 GHz, and multifrequency feature set improves the accuracy to 92%.

@InProceedings{Savazzi_19_icassp,
  author    = {V. Rampa, S. Savazzi, M. D'Amico, G. G. Gentili},
  booktitle = {Proc. of Topical Conference on Antennas and Propagation in Wireless Communications (APWC)},
  title     = {Dual target body model for device-free localization applications},
  year      = {2019},
  abstract  = {A dual-target model for quantitative evaluation of the influence of two people standing or moving in the surroundings of a radio link is presented here. This physical model, based on the scalar diffraction theory, is able to predict the effects of the attenuation of the electromagnetic (EM) wavefield generated by a RF transmitter caused by the presence of two people standing or moving in the area covered by a radio link. This model allows to relate, for each link, the RSS measurements at the receiver to the position, size and orientation, of two people located in the link area. Unlike complex EM frameworks that cannot be adopted for real- or near real-time applications, the proposed model can be employed for crowd sensing, occupancy estimation and people counting applications for both indoor and outdoor scenarios. In addition, it paves the way to a complete multi-target body model. This novel model toolkit gets over the restrictions of existing simplified multi-body models for DFL applications that exploit linear superposition of single-body effects. The proposed tool is based on the dual knife-edge approach and on a simplified EM body model, but yet effective for DFL applications. Finally, the experimental part shows some preliminary results on the dual-target applications by exploiting RF measurements obtained with WiFi-compliant radio devices working in the 5 GHz band.},
organization={IEEE},
  project = {radiosense}
  }

  

A dual-target model for quantitative evaluation of the influence of two people standing or moving in the surroundings of a radio link is presented here. This physical model, based on the scalar diffraction theory, is able to predict the effects of the attenuation of the electromagnetic (EM) wavefield generated by a RF transmitter caused by the presence of two people standing or moving in the area covered by a radio link. This model allows to relate, for each link, the RSS measurements at the receiver to the position, size and orientation, of two people located in the link area. Unlike complex EM frameworks that cannot be adopted for real- or near real-time applications, the proposed model can be employed for crowd sensing, occupancy estimation and people counting applications for both indoor and outdoor scenarios. In addition, it paves the way to a complete multi-target body model. This novel model toolkit gets over the restrictions of existing simplified multi-body models for DFL applications that exploit linear superposition of single-body effects. The proposed tool is based on the dual knife-edge approach and on a simplified EM body model, but yet effective for DFL applications. Finally, the experimental part shows some preliminary results on the dual-target applications by exploiting RF measurements obtained with WiFi-compliant radio devices working in the 5 GHz band.

@inproceedings{savazzi2019radiosense,
  title={RadioSense: Wireless Big Data for Collaborative Robotics in Smart Factory},
  author={Savazzi, Stefano and Rampa, Vittorio and Vicentini, Federico and Nicoli, Monica Barbara},
  booktitle={Ital-IA Convegno Nazionale CINI sull'Intelligenza Artificiale},
  pages={1--2},
    project = {radiosense},
  year={2019}
}

@inproceedings{savazzi2019motion,
  title={Motion discrimination by ambient cellular signals: machine learning and computing tools},
  author={Savazzi, Stefano and Brondolin, Rolando and Rampa, Vittorio and Santambrogio, Marco and Spagnolini, Umberto},
  booktitle={2019 IEEE 5th World Forum on Internet of Things (WF-IoT)},
  pages={448--453},
  year={2019},
   project = {radiosense},
  organization={IEEE}
}

@inproceedings{rampa2019opportunistic,
  title={Opportunistic sensing in beyond-5G networks: The opportunities of transformative computing},
  author={Rampa, Vittorio and Savazzi, Stefano and Malandrino, Francesco},
  booktitle={Proc. 5G Italy Book Multiperspective View 5G},
  pages={461--475},
   project = {radiosense},
  year={2019}
}

@InProceedings{Savazzi_19_icassp,
  author    = {S. Kianoush and S. Savazzi and V. Rampa},
  booktitle = {Proc. of International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  title     = {Passive Detection and Discrimination of Body Movements in the sub-THz Band: A Case Study},
  year      = {2019},
  abstract  = {Passive radio sensing technique is a well established research topic where radio-frequency (RF) devices are used as real-time virtual probes that are able to detect the presence and the movement(s) of one or more (non instrumented) subjects. However, radio sensing methods usually employ frequencies in the unlicensed 2.4-5.0 GHz bands where multipath effects strongly limit their accuracy, thus reducing their wide acceptance. On the contrary, sub-terahertz (sub-THz) radiation, due to its very short wavelength and reduced multipath effects, is well suited for high-resolution body occupancy detection and vision applications. In this paper, for the first time, we adopt radio devices emitting in the 100 GHz band to process an image of the environment for body motion discrimination inside a workspace area. Movement detection is based on the real-time analysis of body-induced signatures that are estimated from sub-THz measurements and then processed by specific neural network-based classifiers. Experimental trials are employed to validate the proposed methods and compare their performances with application to industrial safety monitoring.},
organization={IEEE},
  project = {radiosense}
  }

Passive radio sensing technique is a well established research topic where radio-frequency (RF) devices are used as real-time virtual probes that are able to detect the presence and the movement(s) of one or more (non instrumented) subjects. However, radio sensing methods usually employ frequencies in the unlicensed 2.4-5.0 GHz bands where multipath effects strongly limit their accuracy, thus reducing their wide acceptance. On the contrary, sub-terahertz (sub-THz) radiation, due to its very short wavelength and reduced multipath effects, is well suited for high-resolution body occupancy detection and vision applications. In this paper, for the first time, we adopt radio devices emitting in the 100 GHz band to process an image of the environment for body motion discrimination inside a workspace area. Movement detection is based on the real-time analysis of body-induced signatures that are estimated from sub-THz measurements and then processed by specific neural network-based classifiers. Experimental trials are employed to validate the proposed methods and compare their performances with application to industrial safety monitoring.

@InProceedings{Savazzi_19_wfiot,
  author    = {S. Savazzi and R. Brondolin and V. Rampa and M. Santambrogio and U. Spagnolini},
  booktitle = {Proc. of 5th World Forum on Internet of Things (WF-IoT)},
  title     = {Motion discrimination by ambient cellular signals: machine learning and computing tools},
  year      = {2019},
  abstract  = {In this paper, we evaluate the capability of built-in cellular radio modems available in several IoT modules to track body motions in their close surroundings, by exploiting the real-time analysis of the omnipresent ambient (or stray) cellular signals. In fact, cellular-based IoT devices constantly monitor and report the received signal quality of the camped and neighbor cells for communication functionality imposed by the cellular standards. These quality signals are extracted and processed here to detect changes in the area nearby. A JSON-REST platform and computing infrastructure have been designed to efficiently store and manipulate in real-time these data samples. Experiments and system validation results are presented for a specific case study where two cellular-enabled devices are converted into sensors, while the cellular signal quality is tracked continuously for classifying body motions.},
organization={IEEE},
  project = {radiosense}
}
  

In this paper, we evaluate the capability of built-in cellular radio modems available in several IoT modules to track body motions in their close surroundings, by exploiting the real-time analysis of the omnipresent ambient (or stray) cellular signals. In fact, cellular-based IoT devices constantly monitor and report the received signal quality of the camped and neighbor cells for communication functionality imposed by the cellular standards. These quality signals are extracted and processed here to detect changes in the area nearby. A JSON-REST platform and computing infrastructure have been designed to efficiently store and manipulate in real-time these data samples. Experiments and system validation results are presented for a specific case study where two cellular-enabled devices are converted into sensors, while the cellular signal quality is tracked continuously for classifying body motions.

@InProceedings{Savazzi_19_Antennas,
  author    = {Stefano Savazzi and Vittorio Rampa and Sanaz Kianoush and D Piazza},
  booktitle = {Proc. of International Symposium on Personal, Indoor and
Mobile Radio Communication (PIMRC)},
  title     = {Pattern Reconfigurable Antennas for Passive Motion Detection: WiFi Test-Bed and First Studies},
  year      = {2019},
  abstract  = {},
organization={IEEE},
  project = {radiosense}
}

@article{Sanaz_2019_mdpi,
author={Sanaz Kianoush and Stefano Savazzi and Vittorio Rampa and Monica Nicoli},
journal={MDPI Sensors},
title={People Counting by Dense WiFi MIMO Networks: Channel Features and Machine Learning Algorithms},
year={2019},
  volume={19},
  number={16},
abstract = {Subject counting systems are extensively used in ambient intelligence applications, such as smart home, smart building and smart retail scenarios. In this paper, we investigate the problem of transforming an unmodified WiFi radio infrastructure into a flexible sensing system for passive subject counting. We first introduce the multi-dimensional channel features that capture the subject presence. Then, we compare Bayesian and neural network based machine learning tools specialized for subject discrimination and counting. Ensemble classification is used to leverage space-frequency diversity and combine learning tools trained with different channel features. A combination of multiple models is shown to improve the counting accuracy. System design is based on a dense network of WiFi devices equipped with multiple antennas. Experimental validation is conducted in an indoor space featuring up to five moving people. Real-time computing and practical solutions for cloud migration are also considered. The proposed approach for passive counting gives detection results with 99% average accuracy.},
project = {radiosense}
}

Subject counting systems are extensively used in ambient intelligence applications, such as smart home, smart building and smart retail scenarios. In this paper, we investigate the problem of transforming an unmodified WiFi radio infrastructure into a flexible sensing system for passive subject counting. We first introduce the multi-dimensional channel features that capture the subject presence. Then, we compare Bayesian and neural network based machine learning tools specialized for subject discrimination and counting. Ensemble classification is used to leverage space-frequency diversity and combine learning tools trained with different channel features. A combination of multiple models is shown to improve the counting accuracy. System design is based on a dense network of WiFi devices equipped with multiple antennas. Experimental validation is conducted in an indoor space featuring up to five moving people. Real-time computing and practical solutions for cloud migration are also considered. The proposed approach for passive counting gives detection results with 99% average accuracy.

@article{niu2019wimorse,
  title={WiMorse: A Contactless Morse Code Text Input System Using Ambient WiFi Signals},
  author={Niu, Kai and Zhang, Fusang and Jiang, Yuhang and Xiong, Jie and Lv, Qin and Zeng, Youwei and Zhang, Daqing},
  journal={IEEE Internet of Things Journal},
  volume={6},
  number={6},
  pages={9993--10008},
  year={2019},
  publisher={IEEE},
  project = {radiosense}
}

@inproceedings{rampa2019physical,
  title={Physical Model-based Calibration for Device-Free Radio Localization and Motion Tracking},
  author={Rampa, Vittorio and Savazzi, Stefano and Kianoush, Sanaz},
  booktitle={2019 IEEE-APS Topical Conference on Antennas and Propagation in Wireless Communications (APWC)},
  pages={353--358},
  year={2019},
  project = {radiosense}
}

@inproceedings{rampa2019dual,
  title={Dual-target body model for device-free localization applications},
  author={Rampa, Vittorio and Savazzi, Stefano and D’Amico, Michele and Gentili, Gian Guido},
  booktitle={2019 IEEE-APS Topical Conference on Antennas and Propagation in Wireless Communications (APWC)},
  pages={181--186},
  year={2019},
  project = {radiosense}
}

@inproceedings{kianoush2019passive,
  title={Passive detection and discrimination of body movements in the sub-THz band: a case study},
  author={Kianoush, Sanaz and Savazzi, Stefano and Rampa, Vittorio},
  booktitle={ICASSP 2019-2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  pages={1597--1601},
  year={2019},
  project = {radiosense}
}

@inproceedings{savazzi2019motion,
  title={Motion discrimination by ambient cellular signals: machine learning and computing tools},
  author={Savazzi, Stefano and Brondolin, Rolando and Rampa, Vittorio and Santambrogio, Marco and Spagnolini, Umberto},
  booktitle={2019 IEEE 5th World Forum on Internet of Things (WF-IoT)},
  pages={448--453},
  year={2019},
  project = {radiosense}
}