About Me

I am a Principal Research Scientist in the Pervasive Systems Department at Nokia Bell Labs in Cambridge, and I lead the Device Forms team. I am also a visitor at the University of Cambridge.

In my team we work on the architectural and algorithmic challenges of building the next-generation wearable systems for human sensing, exploiting ultra-low-power machine learning techniques. Some of the topics we explore include earable computing (esense.io), batteryless sensing, ultra-low power machine learning and acoustic sensing.

I am an experimental computer scientist driven by solving important problems with practical solutions, hence, my work always involves building end-to-end HW and SW systems and evaluating them in real-world deployments. I strongly believe in an interdisciplinary approach to research and I am always looking for opportunities to contribute to the development of new technologies. I contribute to the research community as a committee member for leading mobile and sensor systems conferences and journals.

I finished my PhD in Computer Science at the University of Cambridge (UK) under the supervision of Prof. Cecilia Mascolo. My PhD focused on wearable devices to study social dynamics. I also hold an MEng and a BEng cum laude in Computer Engineering from the University of Bologna (Italy).

During my studies, I have worked on a number of projects in diverse industries, including wireless sensor networks at the ABB Corporate Research Center, privacy issues in the context of the Internet of Things at Nokia Bell Labs, and Internet of Things applications at ARM. I was also a Research Assistant at the University of Cambridge working on optimising drone surveying flight paths using onboard context sensing and computation.

Selected Projects

Adaptive Intelligence for Batteryless Sensors Using Software-Accelerated Tsetlin Machines

ACM SenSys 2022

Tsetlin Machine (TM) is a new machine learning algorithm that encodes propositional logic into learning automata—a set of logical expressions composed of boolean input features—to recognise patterns. TM is a promising candidate for embedding intelligence into tiny batteryless sensors with the potential to address two critical challenges: computing under resource constraints and demand for dynamic adaptation to the unpredictable nature of harvested energy. However, its structural model complexity manifests in two conflicting issues: large memory footprint and long latency. This work addresses these shortcomings by proposing adaptive compression techniques exploiting the inherent redundancies observed in trained models. Through dynamically scaling the computational complexity based on available energy, our techniques significantly reduce the memory footprint and speed up the runtime execution.

Ultra-low Power DNN Accelerators for IoT: Resource Characterization of the MAX78000

ACM AIChallengeIoT 2022

Best Paper Award

The development of edge devices with dedicated hardware accelerators has pushed the deployment and inference of Deep Neural Networks closer to users and real-world sensory systems than ever before. Recently, a further subset of these devices has emerged: ultra-low power DNN accelerators. These microcontrollers possess a dedicated hardware accelerator and are able to operate with only 𝜇J’s of energy. In this work, we take a close look at one such device: the MAX78000 by Maxim Integrated. We characterize the device’s performance by running five DNN models of various sizes and architectures, and analyze its operational latency, power consumption, and memory footprint. In doing so, we offer empirical insights and guidelines for practitioners and researchers interested in deploying DNN models on the MAX78000 platform for ultra-low power sensory tasks.

Non-Invasive Blood Pressure Monitoring with Multi-Modal In-Ear Sensing

IEEE ICASSP 2022

Continuous blood pressure monitoring is the key to mitigate significant risks for stroke, heart failure and coronary artery disease. Current gold-standard blood pressure devices cause discomfort and interfere with users’ activities. In this project we build an earable system which unobtrusively monitors users’ blood pressure from the ear. We propose a cuffless measurement technique based on the vascular transit time which utilises the time difference between the S1 heart sound and the PPG upstroke in one pulse cycle. We develop a multimodal sensing hardware and processing pipeline and we evaluate it with 10 participants showing average errors of 4.07 mmHg for systolic and 5.61 mmHg for diastolic blood pressure.

In-Ear PPG for Vital Signs

IEEE Pervasive Computing 2021

We report a systematic characterization of in-ear photoplethysmography (PPG) in measuring vital signs: heart rate, heart rate variability, blood oxygen saturation, and respiration rate. We explore in-ear PPG inaccuracies stemming from different sensor placements and motion-induced artifacts. We observe statistically significant differences across sensor placements and between artifact types. Our results suggest that in-ear PPG is reasonably accurate in detecting vital signs but demands careful mechanical design and signal processing treatment.

ePerceptive—Energy Reactive Embedded Intelligence for Batteryless Sensors

ACM SenSys 2020

This work presents the design and implementation of ePerceptive: a novel framework for best-effort embedded intelligence, i.e., inference fidelity which varies in proportion to the instantaneous energy supplied. We report the manifestation of ePerceptive in designing batteryless cameras and microphones built with TI MSP430 MCU and off-the-shelf RF and solar energy harvesters. Our evaluation with multiple vision and acoustic workloads suggest that the dynamic adaptation of ePerceptive can increase the inference throughput by up to 80% compared to a static baseline while ensuring a maximum accuracy drop of less than 6%.

Automatic Smile and Frown Recognition with Kinetic Earables

ACM Augmented Human 2019

In this work, we introduce inertial signals obtained from an earable placed in the ear canal as a new compelling sensing modality for recognising two key facial expressions: smile and frown. Borrowing principles from Facial Action Coding Systems, we first demonstrate that an inertial measurement unit of an earable can capture facial muscle deformation activated by a set of temporal microexpressions. Building on these observations, we then present three different learning schemes - shallow models with statistical features, hidden Markov model, and deep neural networks to automatically recognise smile and frown expressions from inertial signals. The results show that in controlled non-conversational settings, we can identify smile and frown with high accuracy (F1 score: 0.85).

Publications

Selected Peer-reviewed Publications

  • Adaptive Intelligence for Batteryless Sensors Using Software-Accelerated Tsetlin Machines
    Abu Bakar, Tousif Rahman, Rishad Shafik, Fahim Kawsar, Alessandro Montanari
    20th Conference on Embedded Networked Sensor Systems (SenSys 2022), November 2022.
    PDF
  • Ultra-low Power DNN Accelerators for IoT: Resource Characterization of the MAX78000 [Best Paper Award]
    Arthur Moss, Hyunjong Lee, Lei Xun, Chulhong Min, Fahim Kawsar, Alessandro Montanari
    4th ACM International Workshop on Challenges in Artificial Intelligence and Machine Learning for Internet of Things (AIChallengeIoT 2022), November 2022.
    Co-located with SenSys 2022.
    PDF
  • A Multidevice and Multimodal Dataset for Human Energy Expenditure Estimation Using Wearable Devices
    Shkurta Gashi, Chulhong Min, Alessandro Montanari, Silvia Santini, Fahim Kawsar
    Nature Scientific Data, September 2022
    Publisher PDF Dataset
  • Non-Invasive Blood Pressure Monitoring with Multi-Modal In-Ear Sensing
    Hoang Truong, Alessandro Montanari, Fahim Kawsar
    47th IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2022), May 2022
    Publisher PDF
  • Logic-based Intelligence for Batteryless Sensors
    Abu Bakar, Tousif Rahman, Alessandro Montanari, Jie Lei, Rishad Shafik, Fahim Kawsar
    23rd International Workshop on Mobile Computing Systems and Applications (HotMobile 2022), March 2022
    Publisher PDF
  • In-Ear PPG for Vital Signs
    Andrea Ferlini, Alessandro Montanari, Chulhong Min, Hongwei Li, Ugo Sassi, Fahim Kawsar
    IEEE Pervasive Computing, December 2021
    Publisher PDF
  • FatigueSet: A Multi-modal Dataset for Modeling Mental Fatigue and Fatigability
    Manasa Kalanadhabhatta, Chulhong Min, Alessandro Montanari, Fahim Kawsar
    EAI International Conference on Pervasive Computing Technology for Healthcare (PervasiveHealth 2021), December 2021.
    PDF Dataset
  • Enabling In-Ear Magnetic Sensing: Automatic and User Transparent Magnetometer Calibration
    Andrea Ferlini, Alessandro Montanari, Andreas Grammenos, Robert Harle, Cecilia Mascolo
    IEEE International Conference on Pervasive Computing and Communications (PerCom 2021), March 2021.
    Publisher PDF
  • ePerceptive: Energy Reactive Embedded Intelligence for Batteryless Sensors
    Alessandro Montanari, Manuja Sharma, Dainius Jenkus, Mohammed Alloulah, Lorena Qendro, Fahim Kawsar
    18th Conference on Embedded Networked Sensor Systems (SenSys 2020), November 2020.
    Publisher PDF
  • A Closer Look at Quality-Aware Runtime Assessment of Sensing Models in Multi-Device Environments
    Chulhong Min, Alessandro Montanari, Akhil Mathur, Fahim Kawsar
    17th ACM Conference on Embedded Networked Sensor Systems (SenSys 2019), November 2019.
    Publisher PDF
  • Resource Characterisation of Personal-scale Sensing Models on Edge Accelerators
    Mattia Antonini, Tran Huy Vu, Chulhong Min, Alessandro Montanari, Akhil Mathur, Fahim Kawsar
    First ACM International Workshop on Challenges in Artificial Intelligence and Machine Learning for Internet of Things (AIChallengeIoT 2019), November 2019.
    Publisher PDF
  • An Early Characterisation of Wearing Variability on Motion Signals for Wearables
    Chulhong Min, Akhil Mathur, Alessandro Montanari, Fahim Kawsar
    International Symposium on Wearable Computers (ISWC 2019), September 2019.
    Publisher PDF
  • Head Motion Tracking Through In-ear Wearables
    Andrea Ferlini, Alessandro Montanari, Cecilia Mascolo, Robert Harle
    1st International Workshop on Earable Computing (EarComp 2019), September 2019.
    Publisher PDF
  • Automatic Smile and Frown Recognition with Kinetic Earables
    Seungchul Lee, Chulhong Min, Alessandro Montanari, Akhil Mathur, Youngjae Chang, Junehwa Song, Fahim Kawsar
    10th Augmented Human International Conference (AH 2019), March 2019.
    Publisher PDF
  • Earables for Personal-Scale Behavior Analytics
    Fahim Kawsar, Chulhong Min, Akhil Mathur, Alessandro Montanari
    IEEE Pervasive Computing, October 2018.
    Publisher PDF
  • Surveying Areas in Developing Regions Through Context Aware Drone Mobility
    Alessandro Montanari, Fredrika Kringberg, Alice Valentini, Cecilia Mascolo, Amanda Prorok
    4th ACM Workshop on Micro Aerial Vehicle Networks, Systems, and Applications (DroNet 2018), June 2018.
    Publisher PDF
  • Measuring Interaction Proxemics with Wearable Light Tags
    Alessandro Montanari, Zhao Tian, Elena Francu, Benjamin Lucas, Brian Jones, Xia Zhou, Cecilia Mascolo
    Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT), March 2018.
    Publisher PDF
  • Detecting Emerging Activity-based Working Traits Through Wearable Technology
    Alessandro Montanari, Cecilia Mascolo, Kerstin Sailer, Sarfraz Nawaz
    Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT), September 2017.
    Publisher PDF
  • A Study of Bluetooth Low Energy Performance for Human Proximity Detection in the Workplace [Best Paper Nominee]
    Alessandro Montanari, Sarfraz Nawaz, Cecilia Mascolo, Kerstin Sailer
    IEEE International Conference on Pervasive Computing and Communications (PerCom 2017), March 2017.
    Publisher PDF

See All Publications...

Patents

  • A System For The Detection Of Artifacts In Bio-Signal Sensors And The Synthesis Of Clean Data For Wearable Devices. {EP 21192515.1}
  • Model modification and deployment. {US 2021/0350280 A1}
  • Mobile Sensor And Nft Based Authentication. {323096-EP}
  • Sensor configuration based on other sensor context determination. {WO 20201/69739 A1}
  • Runtime assessment of sensors. {WO 2021/053444 A1}
  • Detection of facial expressions. {WO 2020/182447 A1}
  • Energy-aware processing system. {WO 2021/180664 A1}
  • Privacy-aware personal data store. {US 2016/0044039 A1}

Mentoring

I find it incredibly stimulating and inspiring mentoring and collaborating with self-motivated students.
At Bell Labs we offer several on-site, paid research internship positions. Click here for more details.
Below are the students I have supervised and collaborated with:

Nokia Bell Labs Interns and Visitors

  • Berken Utku Demirel (ETH) — In-ear sensing for hearing characterisation and enhancement. (2022)
  • Kayla-Jade Butkow (University of Cambridge) — In-ear Sensing for physiological intervention. (2022)
  • Lei Xun (University of Southampton) — Ultra-low Power DNN Accelerators for IoT: Resource Characterization of the MAX78000 (2022)
  • Arthur Moss (Newcastle University) — Ultra-low Power DNN Accelerators for IoT: Resource Characterization of the MAX78000 (2022)
  • Abu Bakar (Northwestern University) — Logic-based Intelligence for Batteryless Sensors and Adaptive Intelligence for Batteryless Sensors Using Software-Accelerated Tsetlin Machines (2021)
  • Tousif Rahman (Newcastle University) — Logic-based Intelligence for Batteryless Sensors and Adaptive Intelligence for Batteryless Sensors Using Software-Accelerated Tsetlin Machines (2021)
  • Ananta Narayanan Balaji (NUS) — In-Ear Sensing (2021)
  • Hoang Truong (University of Colorado Boulder) — Non-Invasive Blood Pressure Monitoring with Multi-Modal In-Ear Sensing (2021)
  • Shkurta Gashi (USI) — A Multidevice and Multimodal Dataset for Human Energy Expenditure Estimation Using Wearable Devices (2021)
  • Manasa Kalanadhabhatta (University of Massachusetts Amherst) — FatigueSet: A Multi-modal Dataset for Modeling Mental Fatigue and Fatigability (2021)
  • Andrea Ferlini (University of Cambridge) — In-Ear PPG for Vital Signs (2020)
  • Manuja Sharma (University of Washington) — ePerceptive—Energy Reactive Embedded Intelligence for Batteryless Sensors (2019)
  • Dainius Jenkus (Newcastle University) — ePerceptive—Energy Reactive Embedded Intelligence for Batteryless Sensors (2019)
  • Mattia Antonini (FBK CREATE-NET and University of Trento) — Resource Characterisation of Personal-scale Sensing Models on Edge Accelerators (2019)
  • Dmitry Ermilov (Skoltech) — Design And Implementation Of Mobile Sensing Applications For Research In Behavioural Understanding (2018)
  • Seungchul Lee (KAIST) — Automatic Smile and Frown Recognition with Kinetic Earables (2018)

Thesis Supervision

  • Alexandru-Mihai Solot — Efficient Deep Learning For Small Object Detection On Embedded Systems
    2019 - University of Cambridge, Master of Philosophy in Advanced Computer Science
  • James Lowenthal — Graph-Based Search for Continuous Space with Application to Path-Planning for Drone-Based Area Surveying
    2018 - University of Cambridge, Computer Science Tripos – Part II
  • Fredrika Kringberg — A Path Planning Approach for Context Aware Autonomous UAVs used for Surveying Areas in Developing Regions
    2017 - KTH, Master of Science Thesis
  • Alice Valentini — Evaluation Of Deep Learning Techniques For Object Detection On Embedded Systems
    2017 - University of Bologna, Computer Science Master Thesis
  • Andrea Bucaletti — Design and Implementation of tuProlog User Interface Following the Java Scripting Engine Specification
    2015 - University of Bologna, Computer Engineering Master Thesis
  • Matteo Librenti — Input Management for tuProlog Graphical Interface in Java and Eclipse
    2013 - University of Bologna, Computer Engineering Bachelor Thesis
  • Alessio Mercurio — Class Loading in Android Applied to the tuProlog Interpreter
    2013 - University of Bologna, Computer Engineering Bachelor Thesis