Emerging Embedded Technologies Lab

EMTECH Lab is directed by professor Fatima Muhammad Anwar at UMass Amherst. Our lab's research is centered around networked embedded systems that interact with the physical world in a safe manner with a human-in-the-loop. We target distributed sensing systems, wearable technologies, Mixed reality (AR/VR), and autonomous systems. Our key design principle is to make distributed embedded systems secure and time-aware, and we strive to create open-source hardware and software artifacts. Work ethics, diversity, and passion drives our research.


  • Dec 2022: PI Fatima was awarded the prestigious NSF CAREER grant
  • Nov 2022: PhD candidate Yasra won Best PhD Forum Presentation Award at ACM SenSys'22
  • April 2022: PhD candidate Yasra chosen as Computing Research Association (CRA) Graduate Fellow for next two years
  • Mar 2022: PhD candidate Yasra earned Honorable Mention in 3MT competition
  • June 2019: PI Fatima featured in ECE UCLA website. Check it out here
  • Mar 2018: PI Fatima selected as Qualcomm Innovation Fellowship Finalist
  • Aug 2017: PI Fatima won best presentation award at N2Women SIGCOMM'17


Trustworthy Time Architecture

Supported by NSF grants,
CAREER: Secure Timing Architecture for Untrusted Edge Systems
CNS Core: Small: Managing Electrical and Thermal Energy in Sustainable Computing System

career's picture

The importance of time security is growing with critical reliance on temporal use cases in emerging IoT applications. A manipulated clock source in edge systems can cause financial loss to humans and fatal consequences to their safety, hence a secure time service is needed that not only provides an accurate time value but also in a timely manner. Any manipulations to either the time value or its delivery results in a wrong sense of elapsed or absolute time. Although there have been advances in isolated domains such as clocksources in trusted execution technologies (TEE), secure time synchronization protocols, and side-channels based resilient high-resolution timers, providing secure time architectures for widespread adoption in edge systems in zero trust environments is a challenge. Our approach to addressing these challenges is centered around diversifying clock sources. Diverse clocks serve as a basis for providing clocks that are integrity-protected, do timely application delivery, and synchronized to other clocks for secure time architectures. Our Processor clocks redesign time stacks within existing trusted hardware architectures for secure and resource-efficient clock source and for immediate adoption. Our Sensor clocks present novel time synchronization interfaces, and Network clocks provide new mechanisms for delay-tolerant synchronization protocols. Together, these diverse clocks assist in secure and extensible design with wide applicability to edge systems

Safe Spaces in Extended Reality

project's picture

A fundamental requirement for Extended Reality (XR) applications is to make the user feel completely immersed in her virtual and/or physical environment safely. The immersiveness relies on multi-modal sensing of user activities and the surrounding environment via commodity sensing devices, such as Head-Mounted Devices (HMD), hand-held controllers, and similar devices. XR systems center on tracking humans and enhancing their experience by designing human-in-the-loop systems. The key challenges most human-in-the-loop systems face: (1) exploitation of security vulnerabilities that directly affect the physical safety of users and (2) loss of cognitive safety because of mismatched human visual and sensory information — caused by a variety of system and development limitations. As a result, immersiveness is a very delicate state and internal or external triggers can easily disturb that, leading to a loss of physical and/or cognitive safety for the user. This project focuses on the fundamental problems in XR platforms and devices that might lead to loss of physical or cognitive safety for a user. Key research thrusts are, (1) exploring XR-specific attack surfaces that a malicious entity can leverage to incur physical harm and (2) quantifying the loss of cognitive safety a user experiences when immersiveness is disturbed. Research contributions are at the intersection of secuity, HCI, and data0driven ML

Fair, Secure, and Efficient Distributed Learning

career's picture

Distributed machine learning techniques such as Federated Learning (FL) preserves privacy via local training at the edge and global model aggregation at the server. Recent research has shown a variety of security, privacy and fairness issues in these techniques. This project focuses on data and model poisoning attacks and defenses on Federated Learning, where incorrect evaluations and assumptions may lead to a false sense of security. The goal is to provide a unified tool that can perform state-of-the-art poisoning attacks on different datasets, distributions, and FL settings for comprehensive security evaluation. Another focus of this project is on performance bias due to data feature heterogeneity in heterogeneous edge devices participating in the federation. The goal is to investigate machine learning techniques such as optimization and weighted empirical risk minimization to build fair applications while maintaining data privacy for edge devices


EMTECH Lab is lead by the following pricinpal investigator and multiple graduate students:

Fatima's picture

Fatima Anwar directs EMTECH Lab at the University of Massachusetts Amherst. Her multidisciplinary research looks into designing trustworthy, time-aware, situational-aware, intelligent, and fair systems for the distributed and resource-constrained edge, specifically for emerging embedded applications in mixed reality, wearable and autonomous technologies. Prior to joining UMass, she was a PhD candidate at UCLA in electrical and computer engineering advised by professor Mani Srivastava. Her dissertation focused on a new perspective of quality-of-time in designing cyber-physical systems. Fatima has a lasting commitment towards increasing participation of diverse groups in computing. She routinely volunteers for outreach activities at the university level i.e. Young Investigator Program, Massenberg STEM Institute, Turing Summer school, Cybersecuirty institute, and gives back to the broader community through programming for Amherst Girl Scouts, Los Angeles Computing Circle (LACC), and Engineering day for Girls

Yasra's pictureYasra Chandio is a fourth-year Ph.D. student at the University of Massachusetts Amherst. Her work focuses on the physical and cognitive safety of mixed reality technology leveraging techniques at the intersection of HCI, data-driven methods and security. She released a Mixed Reality dataset for pose estimation. In 2022, she was a finalist at UMASS 3MT and won Best Presentation award at ACM SenSys PhD Forum. She is also a CRA-E graduate fellow, Grace Hopper scholar, CRA-WP grad cohort scholar, and Google CSRMP scholar

Adeel's pictureAdeel Nasrullah is a fourth-year Ph.D. student at the University of Massachusetts Amherst. He designs secure and lightweight timing services for cyber physical systems at the Edge. His research leverages signal processing, applied machine learning and cyber physical system design using trusted execution environments

Khotso's pictureKhotso Selialia is a a second year PhD student at the University of Massachusetts Amherst working on eliminating performance bias in distributed learning for emerging embedded technologies. Particularly, he is investigating why biases occur and the ways to introduce fariness when integrating Federated Learning in heterogeneous edge devices. Prior to joining UMass, Khotso earned his master's from Carnegie Mellon University Africa

Momin's pictureMomin Khan is a a second year PhD student at the University of Massachusetts Amherst working on the security and privacy of distributed machine learning at the edge. His work focuses on model poisoning attacks on model updates for federated and split learning techniques

Selected Publications

Refer to google scholar for all publications

  • Universal Timestamping with Ambient Sensing[link]
    A Nasrullah, MA Khan, FM Anwar

    IEEE International Conference on Sensing, Communication, and Networking, SECON 2022

  • LTE NFV Rollback Recovery[link]
    MT Raza, Z Tan, A Tufail, FM Anwar

    IEEE Transactions on Network and Service Management, 2022

  • On Key Reinstallation Attacks over 4G LTE Control-Plane: Feasibility and Negative Impact[link]
    MT Raza, Y Guo, S Lu, FM Anwar

    Annual Computer Security Applications Conference, ACSAC 2021

  • Sim2real transfer for deep reinforcement learning with stochastic state transition delays[link]
    SS Sandha, L Garcia, B Balaji, FM Anwar, M Srivastava

    Conference on Robot Learning, CoRL 2021

  • FERRET: Fall-back to LTE Microservices for Low Latency Data Access [link]
    MT Raza, FM Anwar, D Kim, KH Kim

    Usenix Hot Topics in Edge Systems, HotEdge 2020

  • Time Awareness in Deep Learning-Based Multimodal Fusion Across Smartphone Platforms[link]
    SS Sandha, J Noor, FM Anwar, M Srivastava

    ACM/IEEE Conference on Internet of Things Design and Implementation, IoTDI 2020

  • A Case for Feedforward Control with Feedback Trim to Mitigate Time Transfer Attacks[link]
    Fatima M. Anwar, Mani Srivastava.

    ACM Transactions on Privacy and Security, TOPS 2020

  • Securing Time in Untrusted Operating Systems with TimeSeal[link]
    Fatima M. Anwar, Luis Garcia, Xi Han, Mani Srivastava.

    IEEE Real Time Systems Symposium, RTSS 2019

  • Applications and Challenges in Securing Time[link]
    Fatima M. Anwar, Mani Srivastava.

    USENIX Workshop on Cyber Security Experimentation and Test, CSET 2019

  • OpenClock: A Testbed for Clock Synchronization Research[link]
    Fatima M. Anwar, Amr Alanwar, Mani Srivastava.

    IEEE Symposium on Precision Clock Synchronization for measurement, control, and communication, ISPCS 2018

  • Stale time is a security threat[link]
    Fatima M. Anwar, Mani Srivastava.

    USENIX Summit on Hot Topics in Security, HotSec 2018

  • Exposing LTE Security Weaknesses at Protocol Inter-Layer, and Inter-Radio Interactions[link]
    Muhammad Taqi Raza, Fatima M. Anwar, Songwu Lu.

    13th International Conference on Security and Privacy in Communication Networks, SecureComm 2017

  • Cyclops: PRU Programming Framework for Precise Timing Applications[link]
    Amr Alanwar, Fatima M. Anwar, Yi-Fan Zhang, Justin Pearson, Joao Hespanha, Mani Srivastava.

    IEEE Symposium on Precision Clock Synchronization for measurement, control, and communication, ISPCS 2017

  • Timeline: An Operating System Abstraction for Time-Aware Applications[link]
    Fatima M. Anwar, Sandeep D'souza, Adwait Dongare, Anthony Rowe, Raj Rajkumar, Mani Srivastava.

    IEEE Real Time Systems Symposium, RTSS 2016


Introduction to Embedded Systems, ECE 231

Taught in Spring Semester

Embedded systems sense, actuate, compute, and communicate to accomplish tasks in domains such as medical, automotive, and industrial controls. In this course, students will learn the fundamentals of using microcontroller-based embedded systems to solve problems in such domains. By the end of the course, students will be able to choose appropriate hardware based on application requirements, execute and optimize programs on simple microcontrollers, and interface these controllers to other subsystems. These topics will provide a basis for upper-level applied courses, including junior and senior design labs.

This course will consist of a combination of in-class lecture/discussion and lab activities. Each student will receive a kit containing an 8-bit and 32-bit microcontroller development board along with electronic parts and subsystems for interfacing and building small embedded systems. The first half of the course will focus on 8-bit microcontrollers (MCUs) while the second half will focus on 32-bit MCUs using the Beagle Bone Black platform.

Advanced Embedded Systems Design, ECE 524/624

Taught in Fall Semester

A research based graduate level course focused on covering key research areas in networked embedded design. This course presents the unique capabilities of embedded technologies, and takes a holistic approach to design end-to-end systems. These systems span various thrusts that cut across both horizontal and vertical architectural layers. Focused horizontal thrusts are, 1) hardware platforms for emerging applications at the edge, 2) software for bare-metal platforms, and embedded OS 3) network based coordination for distributed entities and 4) cloud-based services for compute-intensive tasks. It also dives into details of vertical thrusts cutting across all layers such as security-aware design, learning based modeling, and resource optimizations in current systems. Finally, the course explores system and security issues that arise with a human-in-the-loop of embedded systems design.

The topics covered in this class equip students with the necessary skill set to research and implement embedded systems from ground up for emerging applications. Students are required to review and critique research papers assigned to them in class, actively participate and lead in discussions, define and implement a semester-long project approved by the instructor, along with presenting key findings and demonstrating the functionality of the project.